Yep, I am
The below resulted in better Timespy scores and lower latency (not by much mind) than when using 3866 / 1933 / CL16
-
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.
5900x Low TimeSpy Score
- Thread starter PureJaz
- Start date
The below resulted in better Timespy scores and lower latency (not by much mind) than when using 3866 / 1933 / CL16
My ram was not configured correctly. With DramCalculator parameters all is fine now: Timespy CPU score 15029
https://ibb.co/5TB0YKk
![a>](http://<a href="https://ibb.co/5TB0YKk"><img src="https://i.ibb.co/nrzG3PB/Dram-calculator.png" alt="Dram-calculator" border="0"></a>)
Sorry to drag this thread back up.
But can you elaborate what about your RAM was configured wrong? DramCalculator values is all greek to me.
I am having very similar performance from my 5900x with Crucial Ballistix 2x 16GB 3600 CL16 RAM.
Bone stock with XMP enabled I was scoring low on everything:
Cinebench R20 : 7700 MC, R23 19000 MC
I enabled PBO2 and put a negative curve of -30 and that improved things slightly:
R20 8200 MC, R23 20700 MC
However my best on 3DMark is like 11,900. Which seems way off the pace.
What I find most frustrating is that just to hit scores that reviewers get stock, I am fiddling with OC settings etc I was hoping not to need to touch.
My temps seem totally fine too using a Dark Rock Pro 4. Max temp under Cinebench single core is about 72, and multi-core its 67~
Single core it seems to hit 4.8-4.9Ghz, and multi-core about 4.3Ghz
Not looking to OC really... just wanting to make sure my 5900x performs as well as it should. (stock ideally!)
All running on a MSI X570 Tomahawk using latest BIOS.
Appreciate any pointers.
Soldato
- Joined
- 30 Mar 2016
- Posts
- 2,884
- Location
- East Lothian
I'm is a similar situation with my 5800X. I only just managed to break Timespy 11000 CPU score by allowing PBO to boost up to 5000 Mhz and applying a -20 to all cores in curve optimiser. Usually scoring around 10500 at stock. I have my RAM running at 3866 / 1933 CL 18 timings. My scores in the CPU-z benchmark are good but seem to be around a 1000 points lower than the norm in the R23 benchmark. I know most folk are running at CL 16 RAM timings which I haven't tried applying yet - I don't know if that's what is a accounting for my lower scores or not?
Last edited:
Highly expect its the RAM causing the lower CPU scores as per a few other posts on here.
This is using a 5800X with default stock (PBO auto) and a 3080 at stock.
RAM is default CL16 / 1800 IF (3600MHz) but running CL14, 1900IF (3800MHz)
https://www.3dmark.com/spy/18076386
Soldato
- Joined
- 30 Mar 2016
- Posts
- 2,884
- Location
- East Lothian
Sounds like it's time to start tweaking the timings though. Where does one normally start with such things? Is there a list of common tried and tested timings available to start from and work up or down from there?
Dam that straight up beats my 5900x + 3080 FE score on every count
So you are manually 'overclocking' your RAM?
I was really hoping to avoid all that by buying 3600CL16 RAM. I just dont have the patience or energy to get into the weeds of RAM timings/voltages and endless stability testing
Really only reason I ran any of these benchmarks was to confirm everything is running as it should. With no intention of overlocking anything. Welp... underclocked CPU... and probably need to underclock the GPU too (for noise!)
Depends on the RAM. Mine is XMP DDR3600, CL16
This thread has lots of useful info - https://www.overclockers.co.uk/foru...ack-ripped-edition-32gb-kits.18885405/page-47
Programs;
ZenTimings_v1.2.1
DRAM-Calculator-for-Ryzen-1.7.3
TestMem5 v0.12 (best configs)
Thaiphoon
Run Thaiphoon to read the memory, export XMP settings them import that into Ryzen DRAM calc. Test default settings. Then load safe and fast configs to see what you can run and change timings in BIOS.
As above though, that thread has lots of useful info
Dam that straight up beats my 5900x + 3080 FE score on every count
So you are manually 'overclocking' your RAM?
I was really hoping to avoid all that by buying 3600CL16 RAM. I just dont have the patience or energy to get into the weeds of RAM timings/voltages and endless stability testing
Really only reason I ran any of these benchmarks was to confirm everything is running as it should. With no intention of overlocking anything. Welp... underclocked CPU... and probably need to underclock the GPU too (for noise!)
Yes.
3600CL16 RAM will of course run fine (run great) but running the same 3600CL16 RAM at 3800CL14 will net better results! (if its stable of course) Some RAM will run faster DDR frequency, some wont even run XMP stable
Trust me, I didnt have the patience either but you'll quickly become addicted to it once you start and before you know it you can reset entire manual RAM timings in BIOS in a minute flat.
Not like there's much outdoor activities, bars, restaurants, holidays or such at the moment
Soldato
- Joined
- 30 Mar 2016
- Posts
- 2,884
- Location
- East Lothian
Thanks for the steer. Appreciate it. I'll take a look...
Thanks for the info!
I already get annoyed enough when I do a BIOS update and need to reset my fan curves
I am guessing that finding stable RAM timings will involve a lot of BIOS resets due to unbootable timings... :|
Ill dip my toe in and take a look at the above programs. See if I can make sense of it all
Cheers
Thanks for the info!
I already get annoyed enough when I do a BIOS update and need to reset my fan curves
I am guessing that finding stable RAM timings will involve a lot of BIOS resets due to unbootable timings... :|
Fans are my bugbear. I really hate them humming / rattling or making much sound at all. I have Noctua 140mm PWM fans but have them all locked at 42% so they dont spin up and down all the time.
It will if you jump in attempting to run DDR3600 at DRR4000 CL14
If you have DDR3600, first increase the voltage to safe limits and try running it at 3800 if IF set to 1900 (so its 1:1 with CPU)
Then slowly work on the timings
Me too!Fans are my bugbear. I really hate them humming / rattling or making much sound at all. I have Noctua 140mm PWM fans but have them all locked at 42% so they dont spin up and down all the time.
It will if you jump in attempting to run DDR3600 at DRR4000 CL14
If you have DDR3600, first increase the voltage to safe limits and try running it at 3800 if IF set to 1900 (so its 1:1 with CPU)
Then slowly work on the timings
I strive for a silent build above all else. I usually set fans and fixed speed regardless, but with how hot 5900x runs, I have gone with a curve this time even for the case fans.
Means when not gaming etc, all the fans settle down to about 550-650rpm.
Under full load they go up to 700-900rpm. So not a huge change...
The 3080 FE I snagged is putting a kybosh on things tho
Louder than Id like. So will look into undervolting it see if that helps. (while still keeping an eye out for a 3080 suprim X )I have low noise adapters on 2 x 140mm CPU fans to limit them to 1200RPM. R20 max temps are 82c using stock PBO defaultsMe too!
I strive for a silent build above all else. I usually set fans and fixed speed regardless, but with how hot 5900x runs, I have gone with a curve this time even for the case fans.
Means when not gaming etc, all the fans settle down to about 550-650rpm.
Under full load they go up to 700-900rpm. So not a huge change...
The 3080 FE I snagged is putting a kybosh on things tho
Idle CPU fans are 400 - 550RPM, 900 - 1200RPM load
3x system fans ~730RPM (42%)
3080 FE idle runs both fans at 1000RPM (30%) with temps around 20c. Cant hear the fans at all.
3080 FE load runs both fans at 2150RPM - 2400RPM (65%) with temps around 60c. Can hear the fans but they are not super loud at all. Certainly not as loud as my 2080Ti. This is runnig the Kombuster stress test though, gaming I'll see ~55c / 1800-1900RPM
Last edited:
I have my Dark rock Pro 4 setup with 40% up to 60 Degress, then slow ramp up to 100% by 85.
Usually goes to about 1200rpm loaded, and 75-80 degress.
Case fans go up to like 900rpm-1000rpm max if CPU hits 85. (only hit that running Furmark + Prime 95 so far...)
The 3080FE I havent touched fan curve yet. Idle its 0rpm, but does get to like 55-57 degress :|
Under 100% load, they hit 2100rpm and like 80degress.
Which is far louder than I am use to and would like. My old MSI GTX970 topped out its fans at like 1450rpm!
Still, the 3080 FE is so awesome looking, and fast as hell.
I think what I really need to do is put the PC bit futher away, maybe under the desk. Notice the noise less then I am sure.
This fan curve is great for the 3080 (and was great on my 2080Ti too)The 3080FE I havent touched fan curve yet. Idle its 0rpm, but does get to like 55-57 degress :|
Under 100% load, they hit 2100rpm and like 80degress.
Which is far louder than I am use to and would like. My old MSI GTX970 topped out its fans at like 1450rpm!
I cant tell the difference between 0% and 30% and prefer the constant 30-50% fan speed, rather than it spinning up and down a lot.
It's never reached 80c, 68c max when stress testing at the same time as CPU and RAM stress testing.
At 100% fan its loud but have only tried that by manually putting it on 100% to see how low I could keep the temps.
Im sure if i put all case, cpu and gpu fans on 100% it wouldnt ever get to 60c even when stress testing
Soldato
- Joined
- 30 Mar 2016
- Posts
- 2,884
- Location
- East Lothian
Gained an extra 1000 points in R23 by enabling performance mode in Windows power settings and by slightly tightening up timings from 18-22-22-41 to 18-19-19-39 at 3866. Also broke 11000 CPU score on Timespy so going in the right direction. Will try to go down Cas 16 when I find some time...
You sure performance mode in windows had any hand in that?
From my limited testing with that on/off is it made no difference :|