BugReporterZ

Not sure if that was meant for Meteor Lake users, but on my own I get this: ``` ./build/corefreq-cli -M Intel Z690 [7A84] Controller #0 Dual Channel Bus Rate 3600...

> @BugReporterZ Hello, > > Can you please pull the latest commits from the [`develop`](https://github.com/cyring/CoreFreq/tree/develop) branch and show the output of `corefreq-cli -k -n -B -n -M` ? I got...

> > I got this: > > Thank you > > Can you tell if the `WR` timing of `24` is correct facing its BIOS setting ? > > Feel...

After pulling and recompiling: ``` ./corefreq-cli -M Intel Z690 [7A84] Controller #0 Dual Channel Bus Rate 3600 MHz Bus Speed 3609 MHz DDR4 Speed 7219 MT/s Cha CL RCDr RCDw...

I'm getting this. ``` ./corefreq-cli -M Intel Z690 [7A84] Controller #0 Dual Channel Bus Rate 3600 MHz Bus Speed 3609 MHz DDR4 Speed 7219 MT/s Cha CL RCDr RCDw RP...

For what it's worth, I have found that Gemma-2-27B quantized to Q6_K often makes mistakes/typos with proper names compared to Gemma-2-8B in Q8_0. I don't think the difference in quantization...

I've observed a similar problem with GGUF files. Standard HF-format files seemed to work correctly. I'm using an NVidia RTX3090.

I also tried with FP16 GGUF files and I can confirm they work with the expected performance on my system. Most people using GGUF files need quantizations, though.

I am observing this error with Mistral Small 3/3.1/3.2 (24B) even though I should have plenty of memory to load the model in 4-bit, having a 24GB GPU. This occurs...

Might be related to https://github.com/unslothai/unsloth/issues/2504 There, an [example was shown](https://github.com/unslothai/unsloth/issues/2504#issuecomment-2866246304) of a 30B parameter model requiring 43GB of memory while the shards were being loaded, and 16.8GB after loading.