Contribute
Register

Low Metal Score at fresh boot with "Pro" SMBIOS, could you try?

Status
Not open for further replies.
Joined
Mar 22, 2011
Messages
234
Motherboard
Gigabyte Z690 Aorus Elite DDR4
CPU
i5-13600KF
Graphics
RX 6600 XT
Hallo!

So I just built an Z690 system, and I'm really happy about it!
I used many of my components from my previous Z490 build, including an RX580 8Gb video card.

Everything works well but I noticed pretty low scores in Geek Bench 5 Metal test, around 37000, where I previously had 52000.

I did a lot of testing and investigation, and thanks to a suggestion I discovered that if I sleep the machine and then wake it the score goes up to the proper RX580 result (well over 50000).

So I tried back on my Z490...where it didn't happen (using iMac20,1). I then tried both iMacPro1,1 and MacPro7,1 SMBIOS on that Z490 and even there I get the same weird behaviour, low scores that go up after a sleep! So it definitely looks like a "Pro" SMBIOS problem!

Could other Z690 users (or even Z490 using iMacPro1,1) try that please, and share your results? Also curious to see if it only happens with an RX580 or even with newer cards!

Many thanks!
 
Please, can someone post results? It should be a simple test! Many thanks!
 
Please, can someone post results? It should be a simple test! Many thanks!
No I get same metal score after sleep no difference this on z690 MacPro7,1 rx 6900 spoofed
 
Please, can someone post results? It should be a simple test! Many thanks!

It's a known apple issue with Polaris GPU.
iMac20,1 does not have this issue.
There's no solution.

Measure real world Apps (FCPX, VideoProc, Word, Excel, Powerpoint, etc, etc) to see if it is a real problem for you as synthetic benchmarks are not what people use all day.
 
It's a known apple issue with Polaris GPU.
iMac20,1 does not have this issue.
There's no solution.

Measure real world Apps (FCPX, VideoProc, Word, Excel, Powerpoint, etc, etc) to see if it is a real problem for you as synthetic benchmarks are not what people use all day.
Thanks for the reply! Other tests gives same results before and after sleep. OpenCL tests are fine. Tried GFX bench Metal, and even there there seems to be no difference in scores before or after sleep (even though they look really low). Can't really know if apps perform better after sleep, no way to visually measure it. FCPX exports bruce X in 8 seconds, even after a fresh boot... so it definitely seems like some bug only related to GB5
 
Thanks for the reply! Other tests gives same results before and after sleep. OpenCL tests are fine. Tried GFX bench Metal, and even there there seems to be no difference in scores before or after sleep (even though they look really low). Can't really know if apps perform better after sleep, no way to visually measure it. FCPX exports bruce X in 8 seconds, even after a fresh boot... so it definitely seems like some bug only related to GB5

"Other" tests do very specific tests. GB does a range of synthetic tests.
That why it's better to do real-life App tests...like the Bruce test or measure time to encode a clip and and ignore GB.
That is, don't get hung up on GB.
It has been reported many times and wherever the problem is, it is not going to get fixed.
 
This sounds similar to a problem I had with RX580 paired with a Ryzen 7 5700X, link below.

Thread 'ASUS Prime A320I-K — Ryzen 7 5700X — RX 580 — Ventura'
https://www.tonymacx86.com/threads/asus-prime-a320i-k-ryzen-7-5700x-rx-580-ventura.321400/

I was getting Metal GB5 scores around 37K when using SMBIOS MacPro7,1. When I changed it to iMac19,1 the Metal scores jumped to 57K.
I see in that post many were getting crazy and blaming Ventura...while a little nap on the machine would "solve" the problem :). Just very curious to know why that happens only on "pro" smbios... and would also really love to see if using real apps shows any real difference before or after sleep. Anyway I mainly use Adobe software for content creation, which uses OpenCL... so I should be fine anyway
 
Status
Not open for further replies.
Back
Top