• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

13900K underperforming by 20% in 3dmark, fine (40k) in Cinebench r23.

Prozn

Prozn

Prozn

Associate
Joined
16 Apr 2015
Posts
274
Morning,

I am comparing my system with a nearly identical friend's PC and our benchmarking is revealing some interesting CPU results - in 3dmark my CPU is underperforming by about 20%.

Specs:

MeFriend
CPU (Diff: K vs KF)13900K13900KF
Mobo (Diff: Gigabyte vs Asus / Z790 vs Z690)Gigabyte Z790 Aorus MasterAsus ROG STRIX Z690-F Gaming Wifi
Memory (Same)32gb G.Skill DDR5 6400Mhz 32-39-39-102-14132gb G.Skill DDR5 6400Mhz 32-39-39-102-141
GPU (not relevant for this test)RTX 4090 Founders EditionGigabyte Aorus RTX 4090 Master
Cooling (temps are similar during tests)360mm AIO (Corsair)360mm AIO (NZXT)
Primary Drive (Same other than size)1Tb Samsung 980 Pro512Gb Samsung 980 Pro

Results:

In Cinebench we score fairly similar, with my system edging a slight advantage: 40,068 vs 38,963 (2.8% better). This is margin of error as my friend was not in full "benchmarking mode" when running Cinebench.

In 3dmark however, CPU scores are: 19,876 vs 23,658 (friend is 19% better). This has been the same over multiple runs. (result link: https://www.3dmark.com/compare/spy/32331542/spy/31887590#)

Friend is running default Asus settings with latest bios, and I was running default Gigabyte settings with latest bios.

Things I have tried:
  • Disabling IGPU
  • Removing AVX offset (Was 5 on Gigabyte auto, now 0)
  • Enabling resizable bar (probably not relevant, but was off by default)
Looking in XTU (just to see values, not used for overclocking), my power limits are removed and everything seems normal (although I am coming from a 9900K, where this was all a lot more simple):

j4MlA2q.png

Does anyone have any ideas what could be causing this? My CPU score looks to be one of the lowest on 3dmark. It is possible I just have a bad chip, but I would expect ~5% perf difference maybe, not nearly 20%!

Thanks!
 
Last edited:
Oddly in 3dmark the peak power usage (CPU package power in hwinfo) is 190W and ~70C. I am not sure if that power usage is normal for 3dmark (seems low?), but there is quite a bit of thermal headroom.

I am hitting ~90C on single pass cinebench runs, pulling 307W - that will slowly increase to 100C throttling (but only after quite a long time running). Peak vcore was 1.368V.
 
With unlimited power limit up to 350W draw is expected in R23 I think - this board came with unlimited PL at stock.

My main concern is why 3dmark is performing so poorly. Once I've figured that out I will apply more "daily driver" PL settings.
 
Timespy doesn't like more than 16 threads. (Solution on amd would be to disable 1 ccd or smt to improve performance) Also running the steam version will lower scores. How much will deend on whatt steam is doing int he background.
 
Last edited:
MadMatty said:
Also running the steam version will lower scores. How much will deend on whatt steam is doing int he background.
Click to expand...

Thanks, I am downloading standalone now and will see how that goes - it seems unlikely it would cause a 20% difference though.

As we are doing apples for apples on 13900K the 16 threads issue should affect both equally I would have thought?
 
Prozn said:
Thanks, I am downloading standalone now and will see how that goes - it seems unlikely it would cause a 20% difference though.

As we are doing apples for apples on 13900K the 16 threads issue should affect both equally I would have thought?
Click to expand...
Windows 11H2 installed ? Had an issue where that caused a 5K loss in cpu score for me.

Time Spy Standard DX 12 Bench.

Scoreboard updated.:)
forums.overclockers.co.uk forums.overclockers.co.uk
 
I'm on AMD so this is just a theory here.... temps and boosting

KF is sold without a gpu, is it possible that your friends pc is able to boost for longer during the test meaning it ends up with a higher cpu score? It might also be down to the cooling too, cooler room, better for the radiator, the nzxt rad/fans perform better than the corsair rad/fans combo....

Disabling the GPU isn't quite the same as 'not having one'

EDIT: another thought... background tasks such as rgb software or some other sort of control software. I know when I used lian li AL120 fans the software could easily take 5-10% cpu just running in the background.
 
Last edited:
momo56 said:
Windows 11H2 installed ? Had an issue where that caused a 5K loss in cpu score for me.

Time Spy Standard DX 12 Bench.

Scoreboard updated.:)
forums.overclockers.co.uk forums.overclockers.co.uk
Click to expand...

11H2 was not installed, although MS has just pushed the update to me today!

arknor said:
do you actually have xmp enabled for your ram
Click to expand...

Yes, definitely on.

lsg1r said:
I'm on AMD so this is just a theory here.... temps and boosting

KF is sold without a gpu, is it possible that your friends pc is able to boost for longer during the test meaning it ends up with a higher cpu score? It might also be down to the cooling too, cooler room, better for the radiator, the nzxt rad/fans perform better than the corsair rad/fans combo....

Disabling the GPU isn't quite the same as 'not having one'

EDIT: another thought... background tasks such as rgb software or some other sort of control software. I know when I used lian li AL120 fans the software could easily take 5-10% cpu just running in the background.
Click to expand...

Yes this could be part of it - I do have a few RGB things running and my room is hotter. Although my CPU temps are similar to his!

Deluxe1 said:
With my 13900k in timespy I get 20335, my friend with his 13900k gets about the same too.
Click to expand...

My friend has re-run 3dmark now following a bios update a week or so ago and his performance has dropped back in line with mine (https://www.3dmark.com/compare/spy/32353402/spy/32331542)! So this has possibly all been a big red herring. It does beg the question how the old bios was able to achieve such a higher score though.
 
Prozn said:
Yes this could be part of it - I do have a few RGB things running and my room is hotter. Although my CPU temps are similar to his!
Click to expand...
Most current cpu's work to a maximum temperature rather than a set maximum frequency. It's entirely possible for your pc and your friends pc to show the same temps but for one to perform lower than the other due to their external circumstances.

As for your friends bios update... a new BIOS update can have more impact than you think, my AMD one gets reset every flipping time meaning I've had to manually configure it after each update.... if the intel one is the same it could be a case of a single setting being missed or, again as with an AMD update I've seen, they change a default setting which in turn makes certain things perform slightly worse.
Not to mention microcode updates for patching things like (although obviously not in this case) spectre can impact speeds too.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom