Soldato
Same here, I've updated 2 systems in that way without issue.
-
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.
***The Official 5900X \5950X owners thread***
- Thread starter easyrider
- Start date
just got a 5900x today and shock at how much cooler it runs than my 5800x at an all core load. but set prime to 8 cores and its same as 5800x temps lol.
can anyone point out what stepping i have? in cpuz under stepping it says 2 not sure if thats B2?
also are those bench results looking ok? iv set ppt to 145w everything else is stock no oc or undervolt.
can anyone point out what stepping i have? in cpuz under stepping it says 2 not sure if thats B2?
also are those bench results looking ok? iv set ppt to 145w everything else is stock no oc or undervolt.
Associate
- Joined
- 3 Feb 2011
- Posts
- 1,213
Seems okay, mine's the same revision, but my single thread score is a bit lower (640~) and multi thread score higher (9600~).
I don't think CPU-Z benchmark really tells you the full picture though.
I don't think CPU-Z benchmark really tells you the full picture though.
iv made some changes and setup manual pbo settings of ppt 165w tdc 114a and edc 140a, benches have gone up some:
scores higher than before in cinebench too. i didnt like the single core voltage spikes going to 1.5v so set edc to 141a now so it limits vid to 1.425v max, single core bench is now lower at 672ish in cpuz but all core is same as screenshot above. may faff with the settings some more. cant get anything on curve optimizer even -5 errors on 2 cores in core cycler test.
one thing thats disappointing is that my handbrake encoding speed is about the same as the old 5800x. encoding cartoons in 1080p with x265 and where the 5800x was pegged at 100% usage the 5900x is at around 70% usage. had to do 2 simultaneous encodes to peg cpu at 100%. guess there is a point were more cores dont help and faster cores are more beneficial.
also when stressing for temps prime95 sfft avx gets the cpu to hit max tdc but ppt pull is around 140w and clocks at 3.7ghz and temps around 78c. but aida stress test ticking cpu and fpu gets the cpu hotter to 86c and has higher power pull at 155w but dont make sense since tdc and edc are both at 100%. aida cpu speed is around 4150mhz during stress. weird since i thought clock speeds would be same on all high workloads that pull the same amount of amperage.
Last edited:
5900x supposed to have base clock of 3.7ghz, but iv noticed at stock settings out of the box when running prime 95 sfft the all core clocks are a lot less than 3.7ghz and more around 3.4ghz. anyone else notice that or is it just me??
Caporegime
That would be very odd. What temps are you seeing? With half decent cooling you should be seeing well over 4ghz.
Last edited:
which torture test are you running in prime95 so I can test ?
temps are around 75c on all 12 core test. put it to test 4 cores and it shoots up straight to 90c throttle point.
this one:
thats after running for a few minutes stays around 3.7ghz , I aint running it at stock I have done core optimizer per core
Last edited:
ahh co will be whats throwing off the readings then. since im running full stock with no co. co will reduce power pull in return that power gets used up again for more boost clock.thats after running for a few minutes stays around 3.7ghz , I aint running it at stock I have done core optimizer per core
are you able to test without co and true stock settings?
cinebench is around 4ghz but thats not as heavy a load as prime. prime hits the tdc hard. raising tdc seems to raise the clocks of prime and power consumption. its odd that at prime running 8 core test temps are higher and ppt is higher too than when running 12 core test which has less ppt and lower temps.
this is cpu at stock after 2mins it mostly stays at around 3.5ghz I think its more to do with you getting higher temps 75c to my 58c so slight drop would seem about right also out of interest what CPU package power are you getting ? my 5900x is the early ones so isnt B2 stepping
Last edited:
Caporegime
Prime must drop all core clocks more than I remember, apologies! 3.4 is still pretty low but I'd be more interested in real world loads than P95.
Last edited:
That's good. So at stock it does go below 3.7ghz in prime so thats normal i guess.
I did try curve optimiser at -10 on all cores and it boosts cinebench score a fair bit but doesnt pass core cycler testing. Even -5 is unstable in core cycler. Dunno if its worth tweaking the weakest cores with co but as it is running stock its great.
Iv done some tweaks now, raised ppt to 165w and tdp to 107a edc is at 141a, this has raised all core clocks some in cinebench and temps level off at around 80c. Although in all my testing with aida, ycruncher, prime etc etc max ppt iv seen is 163w.
Handbrake encoder with the cartoons i encode only hits and holds 100% cpu usage if i do 2 parallel encodes. Guess it's the limitation of 1080p vids. 4k would probably hit cpu more
I did try curve optimiser at -10 on all cores and it boosts cinebench score a fair bit but doesnt pass core cycler testing. Even -5 is unstable in core cycler. Dunno if its worth tweaking the weakest cores with co but as it is running stock its great.
Iv done some tweaks now, raised ppt to 165w and tdp to 107a edc is at 141a, this has raised all core clocks some in cinebench and temps level off at around 80c. Although in all my testing with aida, ycruncher, prime etc etc max ppt iv seen is 163w.
Handbrake encoder with the cartoons i encode only hits and holds 100% cpu usage if i do 2 parallel encodes. Guess it's the limitation of 1080p vids. 4k would probably hit cpu more
I did first do all core curve optimiser but was never satisfied and slowly having to lower it because of random crashing idle etc, So thought I put in effort and do per core and using core cycler to test everytime core failed I lowered it till I started seeing no errors was time consuming , I didnt mess about with limits etc I could increase peformance but will just use more power and higher temps, Im satisfied with not adding any extra power over stock while increasing performance over stock
this is what I ended up with and its been stable for maybe 6+ months
Last edited:
The cores that boost the highest can't UV as much