3 ms·

This is very helpful too: https://www.canirun.ai/ https://www.canirun.ai/

embedding-shape 5mo ago
Love that it defaults to the GPU being "NVIDIA GeForce 8800 GTX", a GPU released in 2006 with ~700MB of VRAM... The estimates seems far off as well, took https://www.canirun.ai/model/gpt-oss-120b https://www.canirun.ai/model/gpt-oss-120b as an example, with a RTX Pro 6000 and every single number is off, and notably misses estimation for the most important quant for GPT-OSS, the MXFP4 variant.
- rafram 5mo ago
  The default for me was M1. I think it tries to guess using WebGPU.
freeCandy 5mo ago
Every browser gives me a different result, I guess I can't blame the site for that. But it should perhaps mention which browser would be the most accurate.
- jdiff 5mo ago
  Memory bandwidth is completely different on any browser from measured results on my M2 Pro machine. Weirdly, the estimated performance levels and even exact product name differ between Chrome and Firefox. Firefox calls it an M2 Pro and overshoots measured memory bandwidth by 40GB/S, Chrome calls it an "Apple M2 Pro" and overshoots by 80.
kilroy123 5mo ago
Yes, I really like this site too, but it's a bit outdated. "39d ago" in AI time is like 1 year outdated info.
- nullsanity 5mo ago
  [dead]
- jdiff 5mo ago
  I don't think this has been the case for at least 39 days. The news is slowing down. The big headlines now, besides unverified marketing claims, are efficiency gains. Which are fantastic, but don't seem to be met with matching performance gains.
schaefer 5mo ago
> canirun.ai I run dgx spark, and the results here are soooo incomplete for my platform that I can’t trust this site (for my usecase).
- nullsanity 5mo ago
  [dead]
lowbloodsugar 5mo ago
Doesn’t have qwen3 coder next. What else is it missing?
gcr 5mo ago
canirun doesn’t work well for MoE models. Qwen’s 35B A3B model works amazing on my M1 Max 32GB, but canirun thinks it barely runs.

Love that it defaults to the GPU being "NVIDIA GeForce 8800 GTX", a GPU released in 2006 with ~700MB of VRAM... The estimates seems far off as well, took https://www.canirun.ai/model/gpt-oss-120b https://www.canirun.ai/model/gpt-oss-120b as an example, with a RTX Pro 6000 and every single number is off, and notably misses estimation for the most important quant for GPT-OSS, the MXFP4 variant.

rafram 5mo ago
The default for me was M1. I think it tries to guess using WebGPU.

freeCandy 5mo ago

Every browser gives me a different result, I guess I can't blame the site for that. But it should perhaps mention which browser would be the most accurate.

jdiff 5mo ago
Memory bandwidth is completely different on any browser from measured results on my M2 Pro machine. Weirdly, the estimated performance levels and even exact product name differ between Chrome and Firefox. Firefox calls it an M2 Pro and overshoots measured memory bandwidth by 40GB/S, Chrome calls it an "Apple M2 Pro" and overshoots by 80.

kilroy123 5mo ago

Yes, I really like this site too, but it's a bit outdated. "39d ago" in AI time is like 1 year outdated info.

nullsanity 5mo ago
[dead]
jdiff 5mo ago
I don't think this has been the case for at least 39 days. The news is slowing down. The big headlines now, besides unverified marketing claims, are efficiency gains. Which are fantastic, but don't seem to be met with matching performance gains.

schaefer 5mo ago

> canirun.ai I run dgx spark, and the results here are soooo incomplete for my platform that I can’t trust this site (for my usecase).

nullsanity 5mo ago
[dead]

lowbloodsugar 5mo ago

Doesn’t have qwen3 coder next. What else is it missing?

gcr 5mo ago

canirun doesn’t work well for MoE models. Qwen’s 35B A3B model works amazing on my M1 Max 32GB, but canirun thinks it barely runs.