{"id":1389,"date":"2023-04-21T23:30:08","date_gmt":"2023-04-21T23:30:08","guid":{"rendered":"http:\/\/tiemensfamily.com\/timoncs\/?p=1389"},"modified":"2023-04-29T00:59:52","modified_gmt":"2023-04-29T00:59:52","slug":"teraflops-comparison","status":"publish","type":"post","link":"https:\/\/tiemensfamily.com\/timoncs\/2023\/04\/21\/teraflops-comparison\/","title":{"rendered":"Teraflops Comparison"},"content":{"rendered":"\n

Documenting some various GPU hardware<\/p>\n\n\n\n

<\/p>\n\n\n\n

Name<\/td>TFLOPs
single precision<\/td>		TFLOPS
tensor perf (FP16)<\/td>		TFLOPS
(FP16-Sparse)<\/td>		Tensor
cores<\/td>		CUDA cores<\/td>RAM<\/td><\/tr>
RTX 3080Ti<\/td>34.1<\/td>136<\/td>273<\/td>320<\/td>10,240<\/td>12 GB<\/td><\/tr>
V100<\/a>
(specs<\/a>)<\/td>		14<\/td>112<\/td><\/td>640<\/td>5,120<\/td>16 GB<\/td><\/tr>
RTX
3070<\/td>		20.31<\/td><\/td><\/td>184<\/td>5,888<\/td>8 GB<\/td><\/tr>
GTX 1080<\/td>8.8<\/td><\/td><\/td><\/td>2,560<\/td>8 GB<\/td><\/tr>
PS5<\/td>10.3<\/td><\/td><\/td><\/td><\/td><\/td><\/tr>
Xbox X<\/td>12.1<\/td><\/td><\/td><\/td><\/td><\/td><\/tr>
PS4<\/td>1.8<\/td><\/td><\/td><\/td><\/td><\/td><\/tr>
Xbox One<\/td>1.4<\/td><\/td><\/td><\/td><\/td><\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

Note that the V100 is used in the AWS p3.2xlarge instance type. The V100 numbers are in general smaller than the 3080Ti, and with the WSL2 tensorflow 2.12 libraries, the 3080Ti out-performs the V100 on the 50,000 epoch test 736 seconds to 928 seconds – here the 3080Ti is 26% faster.) (Caveat – extremely small test set – only my ml-style-transfer<\/a> code.)<\/p>\n\n\n\n

(Using the “Windows Native tensorflow 2.11” libraries, the V100 out-performed the 3080Ti on the 50,000 epoch test 928 seconds to 1063 seconds – here the V100 is 12% faster).<\/p>\n\n\n\n

It looks like the p3.2xlarge has been around since late 2017<\/a>. It started at $3.06\/hour, and is still the same price today (2023\/Apr). The V100 prices seems to have dropped from $6,000<\/a> in 2019 to $3,500 today.<\/p>\n\n\n\n

Node Replacement Factor (NRF) – nvidia documentation<\/a><\/p>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

Documenting some various GPU hardware Name TFLOPssingle precision TFLOPStensor perf (FP16) TFLOPS(FP16-Sparse) Tensorcores CUDA cores RAM RTX 3080Ti 34.1 136 273 320 10,240 12 GB V100 (specs) 14 112 640 5,120 16 GB RTX3070 20.31 184 5,888 8 GB GTX … Continue reading →<\/span><\/a><\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[1],"tags":[],"_links":{"self":[{"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/posts\/1389"}],"collection":[{"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/comments?post=1389"}],"version-history":[{"count":4,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/posts\/1389\/revisions"}],"predecessor-version":[{"id":1402,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/posts\/1389\/revisions\/1402"}],"wp:attachment":[{"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/media?parent=1389"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/categories?post=1389"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/tiemensfamily.com\/timoncs\/wp-json\/wp\/v2\/tags?post=1389"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}