Question Really weird Passmark 3D Results, has anyone come across this before?

[Avro Arrow]

I'm usually here answering questions, not asking them but I've always been more of a hardware guy than a software guy and this has me completely puzzled.

This is by no means some kind of emergency, it's just something that has piqued my curiosity so please, don't drive yourselves nuts trying to find an answer because this isn't having a negative effect on my gaming experience. It's just a "WTF is up with that?" head-scratcher-type of question.

I'll start with my system specs (they're in my sig but they're all part numbers so I'll just write their names out):

CPU: AMD Ryzen 7 5800 X3D
MOBO: ASRock X570 Pro4
RAM: 16GB Team Vulcan DDR4-3200, 16GB AData XPG DDR4-3200 (all 32GB running at 3.2GHz as per the XMP Profile in my BIOS)
GPU: ASRock Radeon RX 7900 XTX Phantom Gaming OC 24GB (Adrenalin 23.7.1, June 26, 2023)
OS SSD: WD Black SN770 500GB PCIe4 NVMe
PSU: EVGA 1000 G2 Supernova (1kW on the single 12V rail, 80+Gold-Certified)

So, we all know the drill. We get an awesome new toy and we want to see what benchmark scores we can get.
Firestrike Ultra: 19 565
Time Spy Extreme: 10 964

I was very impressed with my 3DMark and Unigine scores, but then I remembered that I have Passmark on my PC and I ran the suite. I usually just use it for the CPU and storage benchmarks and I was reminded as to why.

The 3D scores are so low that it's just plain stupid (except for DX11 and GPU Compute):

DX9: 243 (85th percentile)
DX10: 132 (81st percentile)
DX11: 319 (95th percentile)
DX12: 44 (58th percentile)
GPU Compute: 17859 (97th percentile)

When running the benchmarks, I noticed that the GPU Utilisation showing in the Adrenalin overlay was giving weird numbers:

DX9: ~65% GPU Utlilisation
DX10: ~70% GPU Utilisation
DX11: 60% when looking at the asteroid which jumps to 100% when looking at the space jellyfish
DX12: ~37% GPU Utlilisation
GPU Compute: 75% (particles) to 100% (fractals)

I had intended to ask about this years ago because it also had really weird numbers on my RX 6800 XT. I forgot about it because I promptly set up my RX 6800 XT in a mining rig and wasn't going to use it for six months anyway. However, now we're talking about what is currently the second-most-powerful video card in the consumer market and it's only scoring in the 58th percentile for DX12? And Passmark's test uses less than 40% of it?

With my 3DMark, Unigine and actual gameplay results being nothing short of spectacular and the drivers being rock-solid (didn't have to change the driver package for the new card) I have to assume that the problem is with Passmark.

I know that it's just a meaningless benchmark but it got me curious as to how this could happen from a benchmarking company that has been around for well over a decade. Has anyone else encountered anything like this and, if so, do you know what's causing it?

EDIT: Unlike some people, I always give the award if someone solves my issue.

 

[Cyberat_88]

1) Benchmarks are like taking tests in school, really not indicative of real world performance.
2) Changing resolution to Higher than your old PC lowers your benchmark scores.
3) All comparisons I have studied, made by professionals, show the DX-12 gain over DX-11 is minimal at best.
4) As for DX-9... OpenGL surpassed DX-9 in performance, no benchmark needed.
I did not purchase Passmark, I cannot tell you if they have even risen past Unigine Haven or Valley and those are not DX-12 capable.

 

[Avro Arrow]

1) Benchmarks are like taking tests in school, really not indicative of real world performance.
2) Changing resolution to Higher than your old PC lowers your benchmark scores.
3) All comparisons I have studied, made by professionals, show the DX-12 gain over DX-11 is minimal at best.
4) As for DX-9... OpenGL surpassed DX-9 in performance, no benchmark needed.
I did not purchase Passmark, I cannot tell you if they have even risen past Unigine Haven or Valley and those are not DX-12 capable.

Passmark is free for 30 days after download. I'm sure it's a glitch in their system because my RX 5700 XT got a higher score in DX12 than my 7900 XTX.

You bothered to answer so you get best answer by default. Cheers!

 

[Order 66]

You bothered to answer so you get best answer by default. Cheers!

lol, I don't think I've ever seen a question with just one answer; a whole bunch with no answers sure but never just one.

 

[Avro Arrow]

lol, I don't think I've ever seen a question with just one answer; a whole bunch with no answers sure but never just one.

Well, that's probably because, as someone who usually answers questions like this, if I don't know the answer, it's probably really obscure. I think that it's just a flaw in their benchmark. I never trusted Passmark's GPU benchmarks anyway.

 

[Order 66]

I would love to be more knowledgeable about PCs and answer questions, but I don't have the means to do extensive troubleshooting and am always afraid of making things worse. For example, I should never have messed with my PC's BIOS since I had no idea what I was doing.

 

[Avro Arrow]

I would love to be more knowledgeable about PCs and answer questions, but I don't have the means to do extensive troubleshooting and am always afraid of making things worse. For example, I should never have messed with my PC's BIOS since I had no idea what I was doing.

Don't feel too bad. I assure you that no tech expert has never majorly screwed something up. It's unfortunately a part of the learning process.

 

[Order 66]

Don't feel too bad. I assure you that no tech expert has never majorly screwed something up. It's unfortunately a part of the learning process.

yes, however, the stakes are somewhat high for me as my current PC is really my only competent gaming PC and if I mess something up I am kind of screwed as I don't feel comfortable taking it apart nor do I really have the means or the knowledge to fix it outside of rebooting the PC or uninstalling display drivers. I would love to learn more, however, I would need some cheap hardware where it doesn't really matter if I mess something up and I would also need someone to guide me.

 

[Avro Arrow]

yes, however, the stakes are somewhat high for me as my current PC is really my only competent gaming PC and if I mess something up I am kind of screwed as I don't feel comfortable taking it apart nor do I really have the means or the knowledge to fix it outside of rebooting the PC or uninstalling display drivers.

When I did my first build, I was 12. I assure you that it was the only PC that I had at the time so I know exactly how you feel.

I would love to learn more, however, I would need some cheap hardware where it doesn't really matter if I mess something up and I would also need someone to guide me.

Buy some bottom-of-the-barrel motherboard and CPU, a cheap-as-dirt used video card and some used RAM. No matter how much you pay for something, everything fits together the same way and you can use YouTube videos to guide you.