[{"data":1,"prerenderedAt":1463},["ShallowReactive",2],{"blog-post-en-\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700":3,"blog-posts-sidebar-en":899},{"id":4,"title":5,"body":6,"categories":881,"date":885,"description":886,"extension":887,"heading":888,"image":889,"meta":890,"navigation":552,"originalUrl":891,"path":892,"seo":893,"slug":894,"socialImage":895,"stem":896,"updated":897,"video":897,"__hash__":898},"blog\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700.md","The Radeon AI PRO R9700 does 3D too: Idol in 16.4 seconds",{"type":7,"value":8,"toc":872},"minimark",[9,19,46,49,54,82,87,94,107,119,122,126,132,137,201,214,232,255,265,269,275,280,369,384,389,395,400,480,493,496,500,523,694,708,726,734,743,754,772,868],[10,11,12,13,18],"p",{},"We've spent plenty of time making the R9700 useful for ",[14,15,17],"a",{"href":16},"\u002Fblog\u002Fpaiton-qwen38-mxfp4-dflash2-r9700","local language models",". But a workstation doesn't have to stop at chat and code. What if the same card could turn a picture into an object you can look around?",[10,20,21,22,26,27,30,31,34,35],{},"That's what ",[23,24,25],"strong",{},"Idol, from Sperid Labs",", does. Give it one image and it generates a 3D Gaussian splat, including a plausible back that the picture never showed. With Paiton, a complete request takes ",[23,28,29],{},"16.4 seconds on one Radeon AI PRO R9700",". Stock Idol on that same card takes 39.2 seconds: about ",[23,32,33],{},"2.4× faster",", with the same weights, 50 steps and guidance setting.",[36,37,38],"sup",{},[14,39,45],{"href":40,"ariaDescribedBy":41,"dataFootnoteRef":43,"id":44},"#user-content-fn-bench",[42],"footnote-label","","user-content-fnref-bench","1",[10,47,48],{},"For context, Sperid Labs separately reports 19.1 seconds on an RTX 4090. That is an encouraging reference, not a matched cross-vendor test. Here's what we measured, what comes out, and how to try it.",[10,50,51],{},[23,52,53],{},"In this article",[55,56,57,64,70,76],"ul",{},[58,59,60],"li",{},[14,61,63],{"href":62},"#one-picture-a-whole-object","From one picture to a turntable",[58,65,66],{},[14,67,69],{"href":68},"#from-392-to-164-seconds","The same-GPU result and quality check",[58,71,72],{},[14,73,75],{"href":74},"#where-the-r9700-lands","Other GPUs and an illustrative price comparison",[58,77,78],{},[14,79,81],{"href":80},"#try-it-on-your-r9700","Run it locally",[83,84,86],"h2",{"id":85},"one-picture-a-whole-object","One picture, a whole object",[10,88,89],{},[90,91],"img",{"alt":92,"src":93},"A farmer character: the input image on the left and a rotating view of the generated 3D Gaussian splat on the right.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fturntable-farmer.gif",[10,95,96],{},[97,98,99,100,106],"em",{},"The header is an illustration; this turntable is a real output from the released Paiton package on one R9700, at 262,144 Gaussians. Left: input generated with FLUX.2 klein 4B. Right: Idol's result from all sides. ",[14,101,105],{"href":102,"rel":103},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002F9ff80088fb0063fe0dc140200c21ebc262594aac\u002Fmodels\u002FIdol\u002FREADME.md",[104],"nofollow","Original example and provenance",".",[10,108,109,110,114,115,118],{},"The output is a standard 3DGS ",[111,112,113],"code",{},".ply",", ",[23,116,117],{},"not a textured polygon mesh",". Open it in a Gaussian-splat viewer to explore the object. The unseen sides are generated, not recovered measurements of the original object. That makes this an interesting tool for visual ideas and previews, rather than a replacement for a measured scan.",[10,120,121],{},"A quick turntable makes it easier to spot where a character or prop works, and where another input image would help.",[83,123,125],{"id":124},"from-392-to-164-seconds","From 39.2 to 16.4 seconds",[10,127,128],{},[90,129],{"alt":130,"src":131},"Complete Idol requests on one R9700: stock optimized 39.22 seconds, Paiton exact 19.45 seconds, and Paiton default 16.44 seconds. Lower is better.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fsame-gpu.webp",[10,133,134],{},[97,135,136],{},"Median of eight images after one warm-up. Each request includes input processing, generation and saving the PLY. One R9700, 262,144 Gaussians, 50 steps, CFG 3.0 and time shift 3.0.",[138,139,140,157],"table",{},[141,142,143],"thead",{},[144,145,146,150,154],"tr",{},[147,148,149],"th",{},"Runtime on one R9700",[147,151,153],{"align":152},"right","Complete request",[147,155,156],{"align":152},"Mean LPIPS versus stock optimized",[158,159,160,172,186],"tbody",{},[144,161,162,166,169],{},[163,164,165],"td",{},"Stock Idol, default optimized mode",[163,167,168],{"align":152},"39.22 s",[163,170,171],{"align":152},"Reference",[144,173,174,180,183],{},[163,175,176,177],{},"Paiton ",[111,178,179],{},"--exact",[163,181,182],{"align":152},"19.45 s",[163,184,185],{"align":152},"0.0027",[144,187,188,191,196],{},[163,189,190],{},"Paiton default",[163,192,193],{"align":152},[23,194,195],{},"16.44 s",[163,197,198],{"align":152},[23,199,200],{},"0.0039",[10,202,203,204,207,208],{},"This is about ",[23,205,206],{},"58% less waiting"," than stock. Downloads, startup and the warm-up are outside the timer. Stock uses idol-sdk's optimized PyTorch path on ROCm, not an intentionally slow baseline. The GPU keeps its stock 300 W limit, with no clock, power or fan changes. That limit is not measured energy consumption.",[36,209,210],{},[14,211,45],{"href":40,"ariaDescribedBy":212,"dataFootnoteRef":43,"id":213},[42],"user-content-fnref-bench-2",[10,215,216,217,220,221,223,224],{},"Most of the work is in the denoiser. Paiton replaces it with a native HIP\u002FC++ runtime for RDNA4 ",[111,218,219],{},"gfx1201","; the surrounding pipeline stays on stock idol-sdk and PyTorch. Matrix products remain fp16. The default switches attention to int8 QK and fp8 PV from step 7; ",[111,222,179],{}," keeps fp16 attention throughout.",[36,225,226],{},[14,227,231],{"href":228,"ariaDescribedBy":229,"dataFootnoteRef":43,"id":230},"#user-content-fn-guide",[42],"user-content-fnref-guide","2",[10,233,234,235,238,239,241,242,244,245,248,249],{},"How much does that change the result? The release compared ",[23,236,237],{},"eight objects from eight views each"," against stock renders. Default's mean LPIPS was ",[23,240,200],{},", versus ",[23,243,185],{}," for exact and ",[23,246,247],{},"0.0026"," for stock optimized versus non-optimized. Lower means closer. Default also cleared the release's worst-image PSNR threshold: 34.63 dB against a 33 dB minimum.",[36,250,251],{},[14,252,45],{"href":40,"ariaDescribedBy":253,"dataFootnoteRef":43,"id":254},[42],"user-content-fnref-bench-3",[10,256,257,258,260,261,264],{},"That's a bounded quality check, not a check of every future output. ",[111,259,179],{}," stays near stock's own implementation differences; it does not promise identical bits or bit-reproducibility between runs. Peak ",[23,262,263],{},"PyTorch-reserved GPU memory"," was 9.6 GiB for both Paiton profiles, versus 10.7 GiB for stock. Those are not whole-device measurements or qualification of smaller GPUs.",[83,266,268],{"id":267},"where-the-r9700-lands","Where the R9700 lands",[10,270,271],{},[90,272],{"alt":273,"src":274},"Published Idol request times across GPUs, with ElioVP's R9700 results distinguished from Sperid Labs' upstream measurements. Lower is better.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fcross-gpu.webp",[10,276,277],{},[97,278,279],{},"R9700 measurements: ElioVP. Other GPUs: Sperid Labs' model card. Both report 50 steps, 262,144 Gaussians and medians over eight images, but these are separate measurements, not one controlled hardware comparison.",[138,281,282,295],{},[141,283,284],{},[144,285,286,289,292],{},[147,287,288],{},"GPU and runtime",[147,290,291],{"align":152},"Request",[147,293,294],{},"Source",[158,296,297,308,318,330,340,350,359],{},[144,298,299,302,305],{},[163,300,301],{},"H100 80 GB, variant unspecified",[163,303,304],{"align":152},"6.7 s",[163,306,307],{},"Sperid Labs",[144,309,310,313,316],{},[163,311,312],{},"RTX 5090",[163,314,315],{"align":152},"14.1 s",[163,317,307],{},[144,319,320,323,327],{},[163,321,322],{},"R9700, Paiton default",[163,324,325],{"align":152},[23,326,195],{},[163,328,329],{},"ElioVP",[144,331,332,335,338],{},[163,333,334],{},"L40S",[163,336,337],{"align":152},"16.6 s",[163,339,307],{},[144,341,342,345,348],{},[163,343,344],{},"RTX 4090",[163,346,347],{"align":152},"19.1 s",[163,349,307],{},[144,351,352,355,357],{},[163,353,354],{},"R9700, stock Idol",[163,356,168],{"align":152},[163,358,329],{},[144,360,361,364,367],{},[163,362,363],{},"RTX 3080 10 GB",[163,365,366],{"align":152},"46.2 s",[163,368,307],{},[10,370,371,372,375,376],{},"At these published timings, the R9700 delivers about ",[23,373,374],{},"86% of the RTX 5090's calculated request throughput",". It sits close to the L40S, although a difference this small across separate setups is not a reliable ranking.",[36,377,378],{},[14,379,383],{"href":380,"ariaDescribedBy":381,"dataFootnoteRef":43,"id":382},"#user-content-fn-upstream",[42],"user-content-fnref-upstream","3",[385,386,388],"h3",{"id":387},"what-about-the-hardware-budget","What about the hardware budget?",[10,390,391],{},[90,392],{"alt":393,"src":394},"Illustrative Idol assets per hour per 1,000 US dollars of GPU purchase price: R9700 115.3, RTX 3080 109.7, RTX 5090 77.4, RTX 4090 39.3, L40S 28.6 and H100 17.3.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fgpu-value.webp",[10,396,397],{},[97,398,399],{},"Calculated as 3,600 ÷ request seconds ÷ assumed GPU price × 1,000. An illustrative GPU-only comparison, not an hour-long throughput test. Host costs, energy and services are excluded.",[138,401,402,415],{},[141,403,404],{},[144,405,406,409,412],{},[147,407,408],{},"GPU",[147,410,411],{"align":152},"Assumed price, USD",[147,413,414],{"align":152},"Calculated assets\u002Fhour per $1,000",[158,416,417,429,439,449,459,469],{},[144,418,419,421,424],{},[163,420,322],{},[163,422,423],{"align":152},"$1,900",[163,425,426],{"align":152},[23,427,428],{},"115.3",[144,430,431,433,436],{},[163,432,363],{},[163,434,435],{"align":152},"$710",[163,437,438],{"align":152},"109.7",[144,440,441,443,446],{},[163,442,312],{},[163,444,445],{"align":152},"$3,300",[163,447,448],{"align":152},"77.4",[144,450,451,453,456],{},[163,452,344],{},[163,454,455],{"align":152},"$4,800",[163,457,458],{"align":152},"39.3",[144,460,461,463,466],{},[163,462,334],{},[163,464,465],{"align":152},"$7,570",[163,467,468],{"align":152},"28.6",[144,470,471,474,477],{},[163,472,473],{},"H100 80 GB",[163,475,476],{"align":152},"$31,000",[163,478,479],{"align":152},"17.3",[10,481,482,483,486,487],{},"These are the published README's November 2026 price assumptions, ",[23,484,485],{},"not independently verified lowest available prices",". The H100 price is for PCIe hardware, while the upstream timing does not specify its variant. Treat that row as especially approximate.",[36,488,489],{},[14,490,231],{"href":228,"ariaDescribedBy":491,"dataFootnoteRef":43,"id":492},[42],"user-content-fnref-guide-2",[10,494,495],{},"Under the $1,900 and $4,800 assumptions, the R9700 costs about 40% of the RTX 4090. Prices can move the value ranking: the RTX 3080 is already close here. Use your actual quote, especially if you already own a suitable GPU.",[83,497,499],{"id":498},"try-it-on-your-r9700","Try it on your R9700",[10,501,502,503,512,513,516,517],{},"You need Linux, Docker, a working AMD driver and the tested R9700. ",[23,504,505,506,511],{},"Accept the ",[14,507,510],{"href":508,"rel":509},"https:\u002F\u002Fhuggingface.co\u002Ftimm\u002Fvit_huge_plus_patch16_dinov3.lvd1689m",[104],"DINOv3 license on Hugging Face"," first",", then supply a token for that account. Keep real tokens private; ",[111,514,515],{},"hf_..."," below is only a placeholder.",[36,518,519],{},[14,520,231],{"href":228,"ariaDescribedBy":521,"dataFootnoteRef":43,"id":522},[42],"user-content-fnref-guide-3",[524,525,529],"pre",{"className":526,"code":527,"language":528,"meta":43,"style":43},"language-bash shiki shiki-themes github-light github-dark","docker pull ghcr.io\u002Feliovp\u002Fpaiton-vllm-plugin:idol-rdna4-v1.0.0\n\nexport HF_TOKEN=hf_...\nmkdir -p \"$HOME\u002Fpaiton-idol\u002Fmodels\" \"$HOME\u002Fpaiton-idol\u002Foutputs\"\ndocker run --rm --device \u002Fdev\u002Fkfd --device \u002Fdev\u002Fdri --group-add \"$(stat -c '%g' \u002Fdev\u002Fkfd)\" \\\n  -e HF_TOKEN -v \"$HOME\u002Fpaiton-idol\u002Fmodels:\u002Fmodels\" -v \"$HOME\u002Fpaiton-idol\u002Foutputs:\u002Foutputs\" -v \"$PWD:\u002Finputs:ro\" \\\n  ghcr.io\u002Feliovp\u002Fpaiton-vllm-plugin:idol-rdna4-v1.0.0 generate \u002Finputs\u002Fphoto.png --mode real\n","bash",[111,530,531,547,554,571,597,637,676],{"__ignoreMap":43},[532,533,536,540,544],"span",{"class":534,"line":535},"line",1,[532,537,539],{"class":538},"sScJk","docker",[532,541,543],{"class":542},"sZZnC"," pull",[532,545,546],{"class":542}," ghcr.io\u002Feliovp\u002Fpaiton-vllm-plugin:idol-rdna4-v1.0.0\n",[532,548,550],{"class":534,"line":549},2,[532,551,553],{"emptyLinePlaceholder":552},true,"\n",[532,555,557,561,565,568],{"class":534,"line":556},3,[532,558,560],{"class":559},"szBVR","export",[532,562,564],{"class":563},"sVt8B"," HF_TOKEN",[532,566,567],{"class":559},"=",[532,569,570],{"class":563},"hf_...\n",[532,572,574,577,581,584,587,590,592,594],{"class":534,"line":573},4,[532,575,576],{"class":538},"mkdir",[532,578,580],{"class":579},"sj4cs"," -p",[532,582,583],{"class":542}," \"",[532,585,586],{"class":563},"$HOME",[532,588,589],{"class":542},"\u002Fpaiton-idol\u002Fmodels\"",[532,591,583],{"class":542},[532,593,586],{"class":563},[532,595,596],{"class":542},"\u002Fpaiton-idol\u002Foutputs\"\n",[532,598,600,602,605,608,611,614,616,619,622,625,628,631,634],{"class":534,"line":599},5,[532,601,539],{"class":538},[532,603,604],{"class":542}," run",[532,606,607],{"class":579}," --rm",[532,609,610],{"class":579}," --device",[532,612,613],{"class":542}," \u002Fdev\u002Fkfd",[532,615,610],{"class":579},[532,617,618],{"class":542}," \u002Fdev\u002Fdri",[532,620,621],{"class":579}," --group-add",[532,623,624],{"class":542}," \"$(",[532,626,627],{"class":579},"stat",[532,629,630],{"class":579}," -c",[532,632,633],{"class":542}," '%g' \u002Fdev\u002Fkfd)\"",[532,635,636],{"class":579}," \\\n",[532,638,640,643,645,648,650,652,655,657,659,661,664,666,668,671,674],{"class":534,"line":639},6,[532,641,642],{"class":579},"  -e",[532,644,564],{"class":542},[532,646,647],{"class":579}," -v",[532,649,583],{"class":542},[532,651,586],{"class":563},[532,653,654],{"class":542},"\u002Fpaiton-idol\u002Fmodels:\u002Fmodels\"",[532,656,647],{"class":579},[532,658,583],{"class":542},[532,660,586],{"class":563},[532,662,663],{"class":542},"\u002Fpaiton-idol\u002Foutputs:\u002Foutputs\"",[532,665,647],{"class":579},[532,667,583],{"class":542},[532,669,670],{"class":563},"$PWD",[532,672,673],{"class":542},":\u002Finputs:ro\"",[532,675,636],{"class":579},[532,677,679,682,685,688,691],{"class":534,"line":678},7,[532,680,681],{"class":542},"  ghcr.io\u002Feliovp\u002Fpaiton-vllm-plugin:idol-rdna4-v1.0.0",[532,683,684],{"class":542}," generate",[532,686,687],{"class":542}," \u002Finputs\u002Fphoto.png",[532,689,690],{"class":579}," --mode",[532,692,693],{"class":542}," real\n",[10,695,696,697,700,701,704,705,707],{},"Put ",[111,698,699],{},"photo.png"," in your current directory. Use ",[111,702,703],{},"--mode synthetic"," for renders or generated images, as in the benchmark. Append ",[111,706,179],{}," to retain fp16 attention. Defaults are 50 steps, CFG 3.0, seed 42 and 262,144 Gaussians.",[10,709,710,711,714,715,717,718,721,722,725],{},"The first run downloads the weights; subsequent runs reuse the model directory. Results appear in ",[111,712,713],{},"$HOME\u002Fpaiton-idol\u002Foutputs",": a ",[111,716,113],{},", the prepared ",[111,719,720],{},"_input.png"," and a ",[111,723,724],{},"_camera.json"," file. For a fixed container reference, the public tag resolved on 6 November 2026 to this registry digest:",[524,727,732],{"className":728,"code":730,"language":731,"meta":43},[729],"language-text","ghcr.io\u002Feliovp\u002Fpaiton-vllm-plugin@sha256:7939b4e0bcacf49bcce51d71892915eda802ad2f3724ef61e06e1bc7a77b4e5d\n","text",[111,733,730],{"__ignoreMap":43},[10,735,736,737,742],{},"Use that reference instead of the tag in both Docker commands to pin the image. The ",[14,738,741],{"href":739,"rel":740},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002F9ff80088fb0063fe0dc140200c21ebc262594aac\u002Fmodels\u002FIdol\u002FREADME.md#run",[104],"public release guide"," documents the launcher and downloads.",[10,744,745,746],{},"Package code and the distributed native runtime are Apache-2.0; Paiton's compiler remains proprietary. Idol and the FLUX.2 klein VAE have Apache-2.0 terms, while DINOv3 retains its separate license. Weights download separately.",[36,747,748],{},[14,749,753],{"href":750,"ariaDescribedBy":751,"dataFootnoteRef":43,"id":752},"#user-content-fn-licenses",[42],"user-content-fnref-licenses","4",[10,755,756,757,761,762,766,767,771],{},"Another useful job for the same local workstation. Explore ",[14,758,760],{"href":759},"\u002Fproducts\u002Fpaiton","Paiton",", our ",[14,763,765],{"href":764},"\u002Fblog\u002Fpaiton-qwen-image-21-radeon-ai-pro-r9700","local image-generation work",", or ",[14,768,770],{"href":769},"\u002Fproducts\u002Fsovereign-ai","Sovereign AI"," for keeping these workflows on infrastructure you control.",[773,774,777,782],"section",{"className":775,"dataFootnotes":43},[776],"footnotes",[83,778,781],{"className":779,"id":42},[780],"sr-only","Footnotes",[783,784,785,816,842,855],"ol",{},[58,786,788,793,794,801,802,801,809],{"id":787},"user-content-fn-bench",[14,789,792],{"href":790,"rel":791},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002F9ff80088fb0063fe0dc140200c21ebc262594aac\u002Fmodels\u002FIdol\u002FBENCHMARKS.md",[104],"Idol release benchmark and quality record",". ",[14,795,800],{"href":796,"ariaLabel":797,"className":798,"dataFootnoteBackref":43},"#user-content-fnref-bench","Back to reference 1",[799],"data-footnote-backref","↩"," ",[14,803,800,807],{"href":804,"ariaLabel":805,"className":806,"dataFootnoteBackref":43},"#user-content-fnref-bench-2","Back to reference 1-2",[799],[36,808,231],{},[14,810,800,814],{"href":811,"ariaLabel":812,"className":813,"dataFootnoteBackref":43},"#user-content-fnref-bench-3","Back to reference 1-3",[799],[36,815,383],{},[58,817,819,793,823,801,828,801,835],{"id":818},"user-content-fn-guide",[14,820,822],{"href":102,"rel":821},[104],"Paiton Idol setup, runtime details and published price assumptions",[14,824,800],{"href":825,"ariaLabel":826,"className":827,"dataFootnoteBackref":43},"#user-content-fnref-guide","Back to reference 2",[799],[14,829,800,833],{"href":830,"ariaLabel":831,"className":832,"dataFootnoteBackref":43},"#user-content-fnref-guide-2","Back to reference 2-2",[799],[36,834,231],{},[14,836,800,840],{"href":837,"ariaLabel":838,"className":839,"dataFootnoteBackref":43},"#user-content-fnref-guide-3","Back to reference 2-3",[799],[36,841,383],{},[58,843,845,793,850],{"id":844},"user-content-fn-upstream",[14,846,849],{"href":847,"rel":848},"https:\u002F\u002Fhuggingface.co\u002Fsperidlabs\u002Fidol\u002Fblob\u002F2731c2291fc65a797c5cb627a5e5526fc1490902\u002FREADME.md#across-gpus",[104],"Sperid Labs Idol model card, reviewed revision",[14,851,800],{"href":852,"ariaLabel":853,"className":854,"dataFootnoteBackref":43},"#user-content-fnref-upstream","Back to reference 3",[799],[58,856,858,793,863],{"id":857},"user-content-fn-licenses",[14,859,862],{"href":860,"rel":861},"https:\u002F\u002Fgithub.com\u002FEliovp-BV\u002Fpaiton-vllm-plugin\u002Fblob\u002F9ff80088fb0063fe0dc140200c21ebc262594aac\u002Fmodels\u002FIdol\u002FTHIRD_PARTY_NOTICES.md",[104],"Runtime boundaries and third-party licenses",[14,864,800],{"href":865,"ariaLabel":866,"className":867,"dataFootnoteBackref":43},"#user-content-fnref-licenses","Back to reference 4",[799],[869,870,871],"style",{},"html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .szBVR, html code.shiki .szBVR{--shiki-default:#D73A49;--shiki-dark:#F97583}html pre.shiki code .sVt8B, html code.shiki .sVt8B{--shiki-default:#24292E;--shiki-dark:#E1E4E8}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":43,"searchDepth":549,"depth":549,"links":873},[874,875,876,879,880],{"id":85,"depth":549,"text":86},{"id":124,"depth":549,"text":125},{"id":267,"depth":549,"text":268,"children":877},[878],{"id":387,"depth":556,"text":388},{"id":498,"depth":549,"text":499},{"id":42,"depth":549,"text":781},[760,882,883,884],"AMD Radeon","Local AI","3D Generation","2026-11-06T09:00:00Z","Idol turns an image into a 3D Gaussian splat in 16.44 seconds on one Radeon AI PRO R9700 with Paiton, versus 39.22 seconds for stock on the same GPU.","md","The R9700 does 3D too. Idol in 16.4 seconds.","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fhero.webp",{},"https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700","\u002Fblog\u002Fpaiton-idol-radeon-ai-pro-r9700",{"title":5,"description":886},"paiton-idol-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-idol\u002Fsocial.webp","blog\u002Fpaiton-idol-radeon-ai-pro-r9700",null,"kCPBl2IRbe0PrvaKwcqKPcKw7AYAE5_8HGMchxfS4N0",[900,902,915,925,934,949,958,967,976,985,997,1007,1019,1028,1042,1051,1063,1093,1104,1126,1144,1163,1181,1198,1215,1226,1242,1257,1269,1278,1286,1301,1313,1324,1335,1345,1358,1368,1381,1392,1402,1413,1422,1434,1445,1454],{"path":892,"title":5,"description":886,"date":885,"slug":894,"image":889,"originalUrl":891,"categories":901},[760,882,883,884],{"path":903,"title":904,"description":905,"date":906,"slug":907,"image":908,"originalUrl":909,"categories":910},"\u002Fblog\u002Fpaiton-deepseek-v41-flash-amd-instinct-mi350p","Deepseek V4.1 Flash on a single MI350P","DeepSeek V4.1 Flash on one 144 GB MI350P: 77.9 tokens\u002Fs for one request and 1,107.6 in a 512-request batch, with Paiton and CPU expert offload.","2026-11-03T12:00:00Z","paiton-deepseek-v41-flash-amd-instinct-mi350p","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-deepseek-v41\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-deepseek-v41-flash-amd-instinct-mi350p",[760,911,912,913,914],"AMD Instinct","AI Inference","DeepSeek","Quantization",{"path":916,"title":917,"description":918,"date":919,"slug":920,"image":921,"originalUrl":922,"categories":923},"\u002Fblog\u002Fpaiton-qwen38-flash-next-amd-instinct-mi350p","Qwen3.8 Flash Next on MI350P: twice the speed | Paiton","Qwen3.8 Flash Next on one AMD MI350P: 2.24× the tested vLLM setup's single-request decode speed and about 2.2 million shared KV-cache tokens with Paiton.","2026-10-30T12:00:00Z","paiton-qwen38-flash-next-amd-instinct-mi350p","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-mi350p\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-amd-instinct-mi350p",[760,911,912,924,914],"Qwen",{"path":926,"title":927,"description":928,"date":929,"slug":930,"image":931,"originalUrl":932,"categories":933},"\u002Fblog\u002Fpaiton-qwen38-3bit-accuracy-prefix-caching-mtp","Qwen3.8 Flash Next: 3-bit near 16-bit accuracy, with MTP prefix caching","Paiton's new Qwen3.8 3-bit weights approach 16-bit task scores on two R9700 cards. Native MTP prefix caching resumes a 200K prompt in 0.68 seconds. Results, limits and setup.","2026-10-24T18:45:00Z","paiton-qwen38-3bit-accuracy-prefix-caching-mtp","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-3bit-mtp\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-3bit-accuracy-prefix-caching-mtp",[760,882,883,924,914],{"path":935,"title":936,"description":937,"date":938,"slug":939,"image":940,"originalUrl":941,"categories":942},"\u002Fblog\u002Fnvidia-dsx-pod-vera-rubin-modular-data-centers","Nvidia confirms the modular DC direction","NVIDIA DSX POD brings Vera Rubin AI factories to modular data centers. ElioVP explains site readiness, cooling and the thinking behind ModFlex.","2026-10-21T13:00:00Z","nvidia-dsx-pod-vera-rubin-modular-data-centers","\u002Fasset\u002Fimages\u002Fblog\u002Fnvidia-dsx-pod-modflex\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fnvidia-dsx-pod-vera-rubin-modular-data-centers",[943,944,945,946,947,948],"Modular DC","AI Infrastructure","ModFlex","NVIDIA","Vera Rubin","Liquid Cooling",{"path":950,"title":951,"description":952,"date":953,"slug":954,"image":955,"originalUrl":956,"categories":957},"\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700","Qwen3.8 Flash Next in 4-bit on 2 × Radeon R9700: tested against 16-bit","Paiton runs Qwen3.8 Flash Next in 4-bit on two R9700 cards: 96.8% GSM8K, 96.3% HumanEval, 189.8 tokens\u002Fs and an optional RAM prefix cache. Accuracy, speed and setup.","2026-10-12T15:00:00Z","paiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-4bit-radeon-ai-pro-r9700",[760,882,883,924,914],{"path":959,"title":960,"description":961,"date":962,"slug":963,"image":964,"originalUrl":965,"categories":966},"\u002Fblog\u002Fpaiton-qwen38-flash-next-radeon-ai-pro-r9700","Qwen3.8 Flash Next on 2 × Radeon AI PRO R9700: 216 tokens\u002Fs, 200K context","Paiton runs Qwen3.8 Flash Next with 3-bit experts on two R9700 cards: 216.3 tokens\u002Fs, 565.1 aggregate tokens\u002Fs and a separate 200K mode. Benchmarks, quality and setup.","2026-10-10T10:30:00Z","paiton-qwen38-flash-next-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-flash-next\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-flash-next-radeon-ai-pro-r9700",[760,882,883,924,914],{"path":968,"title":969,"description":970,"date":971,"slug":972,"image":973,"originalUrl":974,"categories":975},"\u002Fblog\u002Fpaiton-qwen38-w3a4-radeon-ai-pro-r9700","Qwen3.8 27B on 1 × Radeon AI PRO R9700: 3-bit weights, 20% faster decode","Qwen3.8 27B on one R9700: 19.9% faster weighted decode, a new 4-bit cache, 200K context and optional vision. Benchmarks and current Paiton setup.","2026-09-27T09:00:00Z","paiton-qwen38-w3a4-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-w3a4\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-w3a4-radeon-ai-pro-r9700",[760,882,883,924,914],{"path":764,"title":977,"description":978,"date":979,"slug":980,"image":981,"originalUrl":982,"categories":983},"Qwen-Image 2.1 on 1 × Radeon AI PRO R9700: 2048×2048 images in 103 seconds","Generate 2048×2048 Qwen-Image 2.1 images locally with 1 × Radeon AI PRO R9700. The released Paiton v1.0.2 container measured 103.29 seconds per warm request, through PNG delivery.","2026-09-23T09:00:00Z","paiton-qwen-image-21-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen-image-21\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen-image-21-radeon-ai-pro-r9700",[760,882,883,984,924],"Image Generation",{"path":16,"title":986,"description":987,"date":988,"slug":989,"image":990,"originalUrl":991,"categories":992},"Qwen3.8: 400.7 tok\u002Fs on R9700 | Paiton","Qwen3.8 on one R9700: 400.7 aggregate tok\u002Fs with ROCm 10 and vLLM 0.29, plus public 200K\u002F220K chat profiles. Benchmarks, limits and launch commands.","2026-09-16T07:30:00Z","paiton-qwen38-mxfp4-dflash2-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-qwen38-mxfp4\u002Fupdate-2026-09-19\u002Fhero.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-mxfp4-dflash2-r9700",[760,882,993,994,995,996],"vLLM","Qwen3.8","Inference Optimization","DFlash2",{"path":998,"title":999,"description":1000,"date":1001,"slug":1002,"image":1003,"originalUrl":1004,"categories":1005},"\u002Fblog\u002Fpaiton-qwen38-neo-gguf-vllm-r9700","Qwen3.8 GGUF in vLLM: Faster Responses on One Radeon","Run the original NEO CODER MAX GGUF in vLLM with Paiton on an R9700. Explore measured latency gains, image input and local deployment.","2026-09-14T07:30:00Z","paiton-qwen38-neo-gguf-vllm-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-neo-gguf\u002F00-hero-neo-gguf-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-neo-gguf-vllm-r9700",[760,882,883,1006,993],"GGUF",{"path":1008,"title":1009,"description":1010,"date":1011,"slug":1012,"image":1013,"originalUrl":1014,"categories":1015},"\u002Fblog\u002Fpaiton-minimax-h3-radeon-ai-pro-r9700","MiniMax H3 on Radeon: 15-Second Video With Native Sound","Paiton generates a 15-second MiniMax H3 video with stereo audio on one Radeon AI PRO R9700 in 5m 33s, with 16.7% lower latency than matched stock.","2026-09-09T07:30:00Z","paiton-minimax-h3-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-minimax-h3\u002F00-featured-minimax-h3-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-minimax-h3-radeon-ai-pro-r9700",[760,882,883,1016,1017,1018],"Video Generation","MiniMax H3","ComfyUI",{"path":1020,"title":1021,"description":1022,"date":1023,"slug":1024,"image":1025,"originalUrl":1020,"categories":1026},"\u002Fblog\u002Fpaiton-flux2-klein-radeon-ai-pro-r9700","Local FLUX.2 klein on Radeon AI PRO R9700: Faster Image Generation with Less VRAM","Paiton generates 1024×1024 FLUX.2 klein images in 1.054 seconds on an R9700: 16.2% lower warm latency and 33.4% lower peak Torch allocation.","2026-09-07T09:00:00","paiton-flux2-klein-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-flux2-klein\u002Ffox-paiton.webp",[760,882,883,984,1027,1018],"FLUX",{"path":1029,"title":1030,"description":1031,"date":1032,"slug":1033,"image":1034,"originalUrl":1035,"categories":1036},"\u002Fblog\u002Fpaiton-ornith15-radeon-ai-pro-r9700","Ornith 1.5 at 44.6 tok\u002Fs on One Radeon AI PRO R9700","Paiton serves Ornith 1.5 35B A3B at 44.63 output tok\u002Fs on one Radeon AI PRO R9700, 27% faster than tuned stock vLLM, with 21.3% lower modeled cost.","2026-09-05T09:00:00","paiton-ornith15-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-ornith15\u002F00-featured-ornith15-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-ornith15-radeon-ai-pro-r9700",[760,1037,882,912,1038,1039,995,1040,993,1041],"Artificial Intelligence","GPU Performance","Inference Latency","Large Language Models","Cost Efficiency",{"path":1043,"title":1044,"description":1045,"date":1046,"slug":1047,"image":1048,"originalUrl":1049,"categories":1050},"\u002Fblog\u002Fpaiton-qwen38-radeon-ai-pro-r9700","Paiton: 21% More Qwen3.8 Throughput on Radeon AI PRO R9700","Paiton serves AMD’s Qwen3.8 27B at 39.77 output tokens\u002Fs on one Radeon AI PRO R9700, delivering 21% more throughput and 17.4% lower modeled cost.","2026-09-04T09:00:00","paiton-qwen38-radeon-ai-pro-r9700","\u002Fasset\u002Fimages\u002Fblog\u002Fpaiton-r9700\u002F00-featured-paiton-r9700.webp","https:\u002F\u002Feliovp.com\u002Fblog\u002Fpaiton-qwen38-radeon-ai-pro-r9700",[760,1037,882,912,1038,1039,995,1040,993,1041],{"path":1052,"title":1053,"description":1054,"date":1055,"slug":1056,"image":1057,"originalUrl":897,"categories":1058},"\u002Fblog\u002Fai-data-center-power-requirements-gpu-per-megawatt","AI Data Center Power Requirements: The GPU\u002FMW Illusion","Why do AI data-center proposals quote different GPU capacities? See how PUE, peak loads, storage, networking and cooling determine deployable compute.","2026-07-27T23:52:00","ai-data-center-power-requirements-gpu-per-megawatt","\u002Fasset\u002Fimages\u002Fblog\u002Fai-data-center-power-requirements-gpu-per-megawatt\u002Fgpu-per-megawatt-illusion.webp",[1059,944,1060,945,1061,1062],"All","Data Centers","HPC","AMD Helios",{"path":1064,"title":1065,"description":1066,"date":1067,"slug":1068,"image":1069,"originalUrl":1070,"categories":1071},"\u002Fblog\u002Fpaiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x","Wan2.2 Video Generation: Paiton on AMD MI355X","Compare Wan2.2-T2V-A14B video generation on AMD MI355X with Paiton and NVIDIA B200 using Diffusers, and explore our diffusion optimization approach.","2026-06-10T14:04:04","paiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonwan2.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-returns-to-its-diffusion-roots-optimizing-wan2-2-t2v-a14b-on-amd-mi355x\u002F",[1059,1037,760,1072,1073,1074,1075,1076,1077,1078,1079,1080,408,1081,1082,1083,1084,1085,1086,1087,760,770,1088,1089,1090,1091,1092],"14B","AMD","B200","Benchmarks","Blackwell","Compute","Diffusion","Eliovp","Generative AI","Hardware","Inference","Instinct","MI355x","NVidia","On-Premise","Optimization","T2V","Text-to-Video","Tuning","Video-Generation","Wan2.2",{"path":1094,"title":1095,"description":1096,"date":1097,"slug":1098,"image":1099,"originalUrl":1100,"categories":1101},"\u002Fblog\u002Ffrom-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure","ElioVP in De Tijd: Chip Optimization and Data Centers","Read about De Tijd's coverage of ElioVP, from its origins in chip optimization to its work on modular data centers and high-density cooling.","2026-02-10T20:48:12","from-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fphysicalnewspaper.webp","https:\u002F\u002Feliovp.com\u002Ffrom-the-attic-to-the-front-page-eliovp-recognized-as-a-pioneer-in-chip-optimization-data-center-infrastructure\u002F",[1059,1037,943,1102,1073,1103,943,1085],"Uncategorized","De Tijd",{"path":1105,"title":1106,"description":1107,"date":1108,"slug":1109,"image":1110,"originalUrl":1111,"categories":1112},"\u002Fblog\u002Fprivacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit","AI Privacy: A Strategic Priority for Benelux Businesses","Explore generative AI privacy risks, trust, data retention and governance, and why Benelux businesses need a strategic approach to secure AI.","2026-01-29T13:51:11","privacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fheaderimage.webp","https:\u002F\u002Feliovp.com\u002Fprivacy-is-geen-it-probleem-meer-het-is-een-strategische-prioriteit\u002F",[1059,1037,1113,1102,1114,1115,1116,1117,1118,1119,1120,1121,1079,1122,1080,1123,1124,1125],"Trending","AI Act","Anthropomorphism","AVG","Benelux","ChatGPT","Cybersecurity","Data Governance","Data Security","GDPR","Microsoft Copilot","Privacy","Shadow AI",{"path":1127,"title":1128,"description":1129,"date":1130,"slug":1131,"image":1132,"originalUrl":1133,"categories":1134},"\u002Fblog\u002Fitsme-bij-ons-is-het-its-not-me-en-dit-is-waarom","Why We Do Not Use itsme: Privacy and Data Sovereignty","Why ElioVP does not use itsme: our assessment of identity metadata, cloud dependence, data sovereignty and authentication risks.","2025-11-27T09:32:14","itsme-bij-ons-is-het-its-not-me-en-dit-is-waarom","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffrontimage.webp","https:\u002F\u002Feliovp.com\u002Fitsme-bij-ons-is-het-its-not-me-en-dit-is-waarom\u002F",[1059,1135,1136,1137,1119,1138,1139,1140,1122,1141,1142,1143,1124],"AWS","Belgian Mobile ID","Cloud Act","Data Sovereignty","Digital Identity","eIDAS","itsme","Liberty Global","MyGov.be",{"path":1145,"title":1146,"description":1147,"date":1148,"slug":1149,"image":1150,"originalUrl":1151,"categories":1152},"\u002Fblog\u002Ffield-report-the-reality-of-building-agentic-ai-in-2025","Field Report. The Reality of Building Agentic AI in 2025","Lessons from building on-premise AI agents in 2025 cover workflow design, observability, model training, hallucinations and GPU memory limits.","2025-11-25T14:03:39","field-report-the-reality-of-building-agentic-ai-in-2025","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffieldreport.webp","https:\u002F\u002Feliovp.com\u002Ffield-report-the-reality-of-building-agentic-ai-in-2025\u002F",[1059,1037,1153,1113,1154,1155,1156,1157,1158,1159,1160,1161,770,1162],"Solutions","Agentic AI","AI Engineering","AI Strategy","Autonomous Agents","Enterprise AI","Local LLM","Model Fine-Tuning","On-Premise AI","VRAM Optimization",{"path":1164,"title":1165,"description":1166,"date":1167,"slug":1168,"image":1169,"originalUrl":1170,"categories":1171},"\u002Fblog\u002Fthe-synthetic-unicorn-bubble","The Synthetic Unicorn Bubble","An analysis of AI neocloud investment risks, examining circular financing, infrastructure claims, contract terms and due diligence.","2025-11-24T19:22:28","the-synthetic-unicorn-bubble","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fsyntheticunicorn.webp","https:\u002F\u002Feliovp.com\u002Fthe-synthetic-unicorn-bubble\u002F",[1059,1037,1113,944,1172,1173,1174,1175,1176,1177,1178,1179,1180],"AI Neocloud","Circular Financing","GPU Cloud","Investment Risks","Startup Valuation","Synthetic Bubble","Tech Analysis","Vaporware","Venture Capital",{"path":1182,"title":1183,"description":1184,"date":1185,"slug":1186,"image":1187,"originalUrl":1188,"categories":1189},"\u002Fblog\u002Fbuilding-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power","NVIDIA GB300 NVL72: A Four-Month Modular Data Center Plan","Explore a modular data center design for NVIDIA GB300 NVL72, covering redundant power, hybrid cooling and a four-month deployment plan.","2025-11-20T14:10:19","building-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fsuperpodmodflexfrontimage.webp","https:\u002F\u002Feliovp.com\u002Fbuilding-the-engine-for-the-ai-race-the-4-month-path-to-nvidia-gb300-nvl72-power\u002F",[1059,943,1102,1190,944,1191,1192,948,1193,1194,1195,1196,1197],"150kW Rack","DLC","High Density","Modular Data Center","NVIDIA Blackwell Ultra","NVIDIA GB300","NVL72","Rapid Deployment",{"path":1199,"title":1200,"description":1201,"date":1202,"slug":1203,"image":1204,"originalUrl":1205,"categories":1206},"\u002Fblog\u002Fwhy-cuda-translation-wont-unlock-amds-real-potential","Why “CUDA” Translation Won’t Unlock AMD’s Real Potential","Why CUDA compatibility is not the same as AMD performance: explore ROCm, HIP, kernel tuning and the case for hardware-specific optimization.","2025-11-12T14:48:37","why-cuda-translation-wont-unlock-amds-real-potential","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fchatgpt-image-nov-11-2025-09_16_10-pm-1.webp","https:\u002F\u002Feliovp.com\u002Fwhy-cuda-translation-wont-unlock-amds-real-potential\u002F",[1059,1037,760,1102,1207,1037,1208,1209,1210,1211,1212,1213,760,1214],"AMD MI300X","CUDA Translation","FP8","GPU Optimization","High Performance Computing","HIP","Kernel Tuning","ROCm",{"path":1216,"title":1217,"description":1218,"date":1219,"slug":1220,"image":1221,"originalUrl":1222,"categories":1223},"\u002Fblog\u002Fpaiton-the-simplest-way-to-supercharge-ai-inference","Paiton: The Simplest Way to Supercharge AI Inference","Learn how Paiton integrates with existing inference stacks, with AMD MI300X benchmark results and performance-per-dollar comparisons.","2025-11-11T10:31:22","paiton-the-simplest-way-to-supercharge-ai-inference","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaiton-powaaah.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-the-simplest-way-to-supercharge-ai-inference\u002F",[1059,1037,760,912,911,1207,1041,1224,995,1213,760,1225,993],"High Throughput","SGLang",{"path":1227,"title":1228,"description":1229,"date":1230,"slug":1231,"image":1232,"originalUrl":1233,"categories":1234},"\u002Fblog\u002Fstop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens","Paiton MoE Benchmarks: MI300X vs H200 and B200","Compare Qwen3-30B-A3B MoE inference with Paiton on MI300X against H200 and B200, including throughput and cost per million tokens.","2025-09-26T13:36:18","stop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fhulkvshulkpaitonwins.webp","https:\u002F\u002Feliovp.com\u002Fstop-overpaying-paiton-mi300x-moe-beats-h200-b200-on-1m-tokens\u002F",[1059,1037,760,1235,1207,1236,995,1237,1238,1239,1240,760,1241],"AI Benchmarks","Cost per Token","Mixture of Experts","MoE","NVIDIA B200","NVIDIA H200","Qwen3",{"path":1243,"title":1244,"description":1245,"date":1246,"slug":1247,"image":1248,"originalUrl":1249,"categories":1250},"\u002Fblog\u002Fagentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en","Local Agentic AI: From Inbox to Action","Local-first AI agents turn email, documents and images into tickets, reports and actions, using models tailored to your data and systems.","2025-09-16T13:09:00","agentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ffrontfotoblog.webp","https:\u002F\u002Feliovp.com\u002Fagentic-ai-but-make-it-local-from-inbox-to-insight-to-action-en\u002F",[1059,1037,1153,1102,1154,1251,1252,1253,1254,1159,1161,770,1255,1256],"Damage Detection","Document Processing","Email Automation","Invoice Extraction","Ticket Automation","Workflow Automation",{"path":1258,"title":1259,"description":1260,"date":1261,"slug":1262,"image":1263,"originalUrl":1264,"categories":1265},"\u002Fblog\u002Fmi300x-fp8-data-parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach","MI300X FP8 Benchmarks: GPU Partitioning with Paiton","Explore Llama 3.1 8B FP8 benchmarks on partitioned MI300X GPUs with Paiton, comparing throughput and latency against NVIDIA H200 and B200.","2025-07-31T13:32:57","mi300x-fp8-data-parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fimage-2-1.webp","https:\u002F\u002Feliovp.com\u002Fmi300x-fp8-data%e2%80%91parallel-benchmarks-8-64-gpus-h200-left-behind-b200-within-reach\u002F",[1059,1037,760,1102,1266,1073,1074,1267,1268,1084,1085,760,993],"AI","H200","MI300X",{"path":1270,"title":1271,"description":1272,"date":1273,"slug":1274,"image":1275,"originalUrl":1276,"categories":1277},"\u002Fblog\u002Fapplicable-ai-for-businesses","Applicable AI for Businesses","Explore ElioVP's approach to local AI for business workflows, including custom model training and automated damage detection for logistics.","2025-07-09T21:35:30","applicable-ai-for-businesses","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fscherm_afbeelding-2025-07-09-om-23.30.17.webp","https:\u002F\u002Feliovp.com\u002Fapplicable-ai-for-businesses\u002F",[1059,1037,1153,1154,1251,1252,1253,1254,1159,1161,770,1255,1256],{"path":1279,"title":1280,"description":1281,"date":1282,"slug":1283,"image":43,"originalUrl":1284,"categories":1285},"\u002Fblog\u002Fintroducing-paitons-free-evaluation-models","Introducing Paiton’s Free Evaluation Models","Test Paiton with free evaluation models for AMD GPUs. Compare text, vision and image generation performance using your own workloads.","2025-07-07T11:26:13","introducing-paitons-free-evaluation-models","https:\u002F\u002Feliovp.com\u002Fintroducing-paitons-free-evaluation-models\u002F",[1059,1037,760],{"path":1287,"title":1288,"description":1289,"date":1290,"slug":1291,"image":1292,"originalUrl":1293,"categories":1294},"\u002Fblog\u002Fpaiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b","Llama 3.1 405B: Faster Startup with Paiton on MI300X","See Paiton benchmarks for Llama 3.1 405B on eight AMD MI300X GPUs, covering model startup, tensor parallelism, throughput and latency.","2025-06-12T20:15:23","paiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fservingscreenshot.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-dramatically-faster-startup-and-performance-for-llama-3-1-405b\u002F",[1059,1037,760,1102,912,1207,1295,1209,1296,1297,1298,760,1299,1300],"Cold Start","Graph Compilation","Llama 3.1 405B","LLM Optimization","Startup Latency","Tensor Parallelism",{"path":1302,"title":1303,"description":1304,"date":1305,"slug":1306,"image":1307,"originalUrl":1308,"categories":1309},"\u002Fblog\u002Fpaiton-fp8-beats-nvidias-h200-on-amds-mi300x","Paiton FP8 Beats NVIDIA’s H200 on AMD’s MI300X","Compare Paiton on AMD MI300X with NVIDIA H200 for Llama 3.1 70B FP8, including throughput, first-token delay and latency across batch sizes.","2025-06-08T19:12:40","paiton-fp8-beats-nvidias-h200-on-amds-mi300x","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fblognewfp8.webp","https:\u002F\u002Feliovp.com\u002Fpaiton-fp8-beats-nvidias-h200-on-amds-mi300x\u002F",[1059,1037,760,1102,1207,1310,1158,1080,1038,1039,1040,1297,1311,1312],"Cold Start Optimization","Model Serving","vLLM Optimization",{"path":1314,"title":1315,"description":1316,"date":1317,"slug":1318,"image":1319,"originalUrl":1320,"categories":1321},"\u002Fblog\u002Fmi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm","MI300X vs H200 vs RX 7900 XTX vs Tenstorrent n300s with vLLM","Compare MI300X, H200, RX 7900 XTX and Tenstorrent n300s on Llama 3 8B with vLLM, including throughput, modeled token costs and hardware limits.","2025-05-09T14:03:58","mi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fcomparisontenstor.webp","https:\u002F\u002Feliovp.com\u002Fmi300x-vs-h200-vs-rx-7900-xtx-vs-tenstorrent-n300s-with-vllm\u002F",[1059,1037,760,1153,1102,1073,1268,1085,1322,1323],"RX7900XTX","tenstorrent",{"path":1325,"title":1326,"description":1327,"date":1328,"slug":1329,"image":1330,"originalUrl":1331,"categories":1332},"\u002Fblog\u002Fclusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights","ClusterP&L: Financial Modeling for GPU Clusters","Explore how ClusterP&L models GPU cluster costs, profitability and investment scenarios, with ROI metrics, risk simulations and exportable reports.","2025-05-03T10:52:22","clusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fcomparisonscenarios.webp","https:\u002F\u002Feliovp.com\u002Fclusterpl-empowering-gpu-cluster-investors-with-real-world-financial-insights\u002F",[1059,1037,943,1153,1074,1267,1333,1085,1334],"MI325x","pnl calculator",{"path":1336,"title":1337,"description":1338,"date":1339,"slug":1340,"image":1341,"originalUrl":1342,"categories":1343},"\u002Fblog\u002Fcranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200","AMD MI300X vs. NVIDIA H200: Qwen3-32B with Paiton","Compare Qwen3-32B benchmarks on Paiton-optimized AMD MI300X and NVIDIA H200, covering throughput, latency and hardware costs.","2025-05-02T21:10:30","cranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002F3ac59a73-2466-4422-b7e5-ef2e4a8ca58e.webp","https:\u002F\u002Feliovp.com\u002Fcranking-out-faster-tokens-for-fewer-dollars-amd-mi300x-vs-nvidia-h200\u002F",[1059,1037,760,1266,1073,1267,1344,1085,760,993],"MI300",{"path":1346,"title":1347,"description":1348,"date":1349,"slug":1350,"image":1351,"originalUrl":1352,"categories":1353},"\u002Fblog\u002Fpower-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw","Modular Data Centers for NVIDIA NVL: 1 to 2 MW","Explore modular data center designs for NVIDIA NVL systems, covering power capacity, liquid cooling, redundancy and deployment planning.","2025-05-02T14:09:59","power-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Feliovp_critical-1mw-pod_rev-2_transparent.webp","https:\u002F\u002Feliovp.com\u002Fpower-meets-precision-high-density-modular-data-center-for-nvidia-nvl-deployments-1-2-mw\u002F",[1059,943,1354,944,1192,1061,948,1193,1355,1356,1196,1357],"1-2MW Data Center","NVIDIA Blackwell","NVIDIA NVL","Precision Cooling",{"path":1359,"title":1360,"description":1361,"date":1362,"slug":1363,"image":1364,"originalUrl":1365,"categories":1366},"\u002Fblog\u002Fexamining-ai-agents-in-the-medical-field-ai-that-speaks-dicom","Examining AI agents in the medical field: AI that speaks DICOM","Explore a local AI agent demo for DICOM workflows, from patient and study retrieval to a comparison of vision models using anonymized medical images.","2025-04-11T14:45:48","examining-ai-agents-in-the-medical-field-ai-that-speaks-dicom","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fhealthcareblog-1.webp","https:\u002F\u002Feliovp.com\u002Fexamining-ai-agents-in-the-medical-field-ai-that-speaks-dicom\u002F",[1059,1037,1153,1102,1266,1073,1367],"Healthcare",{"path":1369,"title":1370,"description":1371,"date":1372,"slug":1373,"image":1374,"originalUrl":1375,"categories":1376},"\u002Fblog\u002Feliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs","U.S. Tariffs and AI Supply Chain Resilience: April 2025","Read ElioVP's April 2025 perspective on U.S. tariffs and supply chain resilience for AI servers, HPC systems and modular data centers.","2025-04-04T10:01:27","eliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ftariffsshipping.webp","https:\u002F\u002Feliovp.com\u002Feliovp-bv-your-trusted-partner-for-supply-chain-resilience-amidst-new-u-s-tariffs\u002F",[1059,1113,1266,1073,1377,1378,1379,1380],"import","Taiwan","Tariffs","Trump",{"path":1382,"title":1383,"description":1384,"date":1385,"slug":1386,"image":1387,"originalUrl":1388,"categories":1389},"\u002Fblog\u002Fwhy-ai-agents-are-the-future","Why AI Agents Are the Future","Explore AI agents for ERP, CRM, finance and customer support, with practical use cases and a path from workflow assessment to pilot and deployment.","2025-03-23T22:06:59","why-ai-agents-are-the-future","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ferp2.jpeg","https:\u002F\u002Feliovp.com\u002Fwhy-ai-agents-are-the-future\u002F",[1059,1037,1153,1266,1390,1391],"AI Agents","ERP",{"path":1393,"title":1394,"description":1395,"date":1396,"slug":1397,"image":1398,"originalUrl":1399,"categories":1400},"\u002Fblog\u002Fthe-rise-of-open-source-ai-model-optimization","The Rise of Open-Source AI Model Optimization","Explore open-source AI optimization trends, from quantization and mixture-of-experts models to hardware-aware tuning, RAG and edge deployment.","2025-03-22T20:59:23","the-rise-of-open-source-ai-model-optimization","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Friseofopensource.jpeg","https:\u002F\u002Feliovp.com\u002Fthe-rise-of-open-source-ai-model-optimization\u002F",[1059,1037,1113,1401,1073,408,1085],"AI news",{"path":1403,"title":1404,"description":1405,"date":1406,"slug":1407,"image":1408,"originalUrl":1409,"categories":1410},"\u002Fblog\u002Fintroducing-our-benchmarking-tool-powered-by-dstack","Introducing Our Benchmarking Tool: Powered by dstack","Explore our dstack-powered tool for reproducible vLLM benchmarks, automated parameter sweeps and performance reports across local and cloud GPUs.","2025-03-20T14:21:59","introducing-our-benchmarking-tool-powered-by-dstack","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fbenchmarktool.jpeg","https:\u002F\u002Feliovp.com\u002Fintroducing-our-benchmarking-tool-powered-by-dstack\u002F",[1059,1037,760,1266,1073,1411,1412,1268,760],"benchmark","LLM",{"path":1414,"title":1415,"description":1416,"date":1417,"slug":1418,"image":1419,"originalUrl":1420,"categories":1421},"\u002Fblog\u002Foptimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200","Optimizing QwQ-32B (by Qwen): AMD MI300X vs. NVIDIA H200","Compare QwQ-32B throughput and latency on AMD MI300X with Paiton and NVIDIA H200, from small batches to higher concurrency.","2025-03-19T21:41:44","optimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaiton4.jpeg","https:\u002F\u002Feliovp.com\u002Foptimizing-qwq-32b-by-qwen-amd-mi300x-vs-nvidia-h200\u002F",[1059,1037,760],{"path":1423,"title":1424,"description":1425,"date":1426,"slug":1427,"image":1428,"originalUrl":1429,"categories":1430},"\u002Fblog\u002Feliovp-featured-on-amd-tech-talk-podcast","Eliovp Featured on AMD “Tech Talk” Podcast","Listen to Elio Van Puyvelde and Jim Greene on AMD's Tech Talk podcast, discussing ElioVP's origins and its AI hardware and software services.","2025-03-19T07:53:39","eliovp-featured-on-amd-tech-talk-podcast","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Ftechtalkjimgreene.jpeg","https:\u002F\u002Feliovp.com\u002Feliovp-featured-on-amd-tech-talk-podcast\u002F",[1059,1073,1431,1432,1433],"Jim Greene","Podcast","Tech Talk",{"path":1435,"title":1436,"description":1437,"date":1438,"slug":1439,"image":1440,"originalUrl":1441,"categories":1442},"\u002Fblog\u002Ffurther-optimizing-amd-powered-inference-with-paiton","Further Optimizing AMD-Powered Inference with Paiton","Explore Paiton's DeepSeek R1 Distill Llama 8B benchmarks on AMD MI300X, focusing on throughput and first-token latency at smaller batch sizes.","2025-03-13T06:18:30","further-optimizing-amd-powered-inference-with-paiton","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost3.webp","https:\u002F\u002Feliovp.com\u002Ffurther-optimizing-amd-powered-inference-with-paiton\u002F",[1059,1037,760,1073,1443,1444,1267,1268,1333,760,993],"Deepseek","H100",{"path":1446,"title":1447,"description":1448,"date":1449,"slug":1450,"image":1451,"originalUrl":1452,"categories":1453},"\u002Fblog\u002Fa-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b","Paiton Benchmarks: DeepSeek R1 Distill Llama 3.1 8B","Compare stock and Paiton-optimized DeepSeek R1 Distill Llama 3.1 8B on AMD MI300X, with throughput and latency benchmarks across batch sizes.","2025-01-31T09:11:02","a-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost2.webp","https:\u002F\u002Feliovp.com\u002Fa-first-look-at-paiton-in-action-deepseek-r1-distill-llama-3-1-8b\u002F",[1059,1037,760,1073,1443,1444,1267,1268,1333,760,993],{"path":1455,"title":1456,"description":1457,"date":1458,"slug":1459,"image":1460,"originalUrl":1461,"categories":1462},"\u002Fblog\u002Fai-model-optimization-with-paiton","AI Model Optimization with Paiton","Learn how Paiton uses model compilation, custom kernels and kernel fusion to optimize AI inference on AMD GPUs.","2025-01-30T19:53:25","ai-model-optimization-with-paiton","\u002Fasset\u002Fimages\u002Fblog\u002Fimported\u002Fpaitonpost1.webp","https:\u002F\u002Feliovp.com\u002Fai-model-optimization-with-paiton\u002F",[1059,1037,760,1073,1444,1267,1268,1333,760,993],1793962343020]