Associate
never run adia latency test, intel burn test, cinebench, geekbench 5 is all i have run, and this one.
8 PACK MEMORY RANGE GROWING: SAY HELLO TO 8 PACK RIPPED EDITION & 32GB KITS!!!
- Thread starter Gibbo
- Start date
never run adia latency test, intel burn test, cinebench, geekbench 5 is all i have run, and this one.
There you go, think it’s a trial. I couldnt find a free one that shows all benchmark stats - https://www.aida64.com/downloads
Associate
Last edited:
Soldato
- Joined
- 28 Sep 2018
- Posts
- 2,532
if you want more memory sensitive tests (cinebench is not):
Geekbench 3
Geekbench 4
X265 benchmark 4K
Y cruncher
Timespy CPU
Associate
Last edited:
Geekbench 5 (5800X)
Single core - 1686
Multi - 11512
@vapor matt Interesting to note your memory speeds at 20GB/s higher at 50 over my 30. 5950X cache related possibly?
My memory copy speed is 52GB/s. Just use benchmark then start to get them instead of using the cache and memory benchmark
I did give your CAS timings and 1800Mhz a try but got memory errors. It does work at 1800Mhz but slightly slackers timings when I tried a few days ago.
Interesting to see that the L3 cache is higher on a 3600 but L1 and L2 is lower...
Single core - 1686
Multi - 11512
@vapor matt Interesting to note your memory speeds at 20GB/s higher at 50 over my 30. 5950X cache related possibly?
My memory copy speed is 52GB/s. Just use benchmark then start to get them instead of using the cache and memory benchmark
I did give your CAS timings and 1800Mhz a try but got memory errors. It does work at 1800Mhz but slightly slackers timings when I tried a few days ago.
Interesting to see that the L3 cache is higher on a 3600 but L1 and L2 is lower...
Associate
Geekbench 5 (5800X)
Single core - 1686
Multi - 11512
@vapor matt Interesting to note your memory speeds at 20GB/s higher at 50 over my 30. 5950X cache related possibly?
My memory copy speed is 52GB/s. Just use benchmark then start to get them instead of using the cache and memory benchmark
I did give your CAS timings and 1800Mhz a try but got memory errors. It does work at 1800Mhz but slightly slackers timings when I tried a few days ago.
Interesting to see that the L3 cache is higher on a 3600 but L1 and L2 is lower...
Your geekbench single core score i would expect to be higher at 350mhz higher clock speed. i can run at 4.7ghz all core but the voltage has to be 1.3v plus and i dont like it to be over that for a 24/7 clock. 4.5ghz 1.27v is a nice 24/7 stable clock all core.
the memory i am using is G-skill trident Z 3600 6-6-6-36-52 samsung B-die but from a year ago. memory speeds might be related to the 5950x to be honest i have no idea if that's the reason or not. One thing is for sure the XMP profile and gigabyte memory training does not load
the best memory settings at all. Mrpils gave me a few settings to try ( thanks Mrpils ) and at the moment i found these to be the best so far. and pleased it can run 3800mhz.
I havent done much in terms of CPU overclocking, just left the PBO on auto. There are a million and one settings for CPU overclocks and voltages. I did have it up to 5050 boost and 4.8Ghz all core with someone elses settings but volts i left on auto so probably not correct and crashed Cinebench R20.
Yeh, MrPils great info needs putting in a sticky really instead of getting burried in a 50 page thread
Associate
the one thing i don't like about PBO is it can give the CPU 1.5v + and that will eventually damage the CPU there are quite a few who have damaged the 3000 series CPU's this way, to the point they wouldn't even run standard clocks any more.
I like to have what i call a safe stable 24/7 overclock, taking a CPU to the brink is ok for 1 off scores but that's all it is a one off score.
Really?? PBO comes as default, which if the default settings are wrecking expensive processors is not good from AMD / motherboard manufacturers. Do you know if there's general voltage settings of 5800X if PBO is not to be used?
Associate
I would set you V core manually to 1.275v and run all core, set V droop to high and run a negative V core offset of say - 0.500v change multiplier to 45 the problem with PBO is all motherboards tend to over volt the CPU to high. then you get spikes where it boosts and then over heats the core and it then down clocks, cools down and boosts again. up and down like a roller coaster.
I didnt like PBO at all, to much voltage when left to the M/B auto settings.
Soldato
Pretty pointless testing 2x8 for comparison. Needs to be 4x8.
I couldnt find setting for vdroop or negstive vcore offset but set the multiplier to 45 and manual vcore. Was stable but all cores only boosted to 4500Mhz (as expected) including single core boost.
Associate
thats right, all core 4500 mhz you can try multi at 46 see how stable it is at 4.6ghz keeping the same V core voltage, but what you should see is all core temp drop from PBO settings. theres not much software that uses single core now, most are multicore.
Using PBO invalidates your warranty because AMD consider it as operating "outside of specifications and in excess of factory settings". I understand them taking this position with people that do manual overclocking because that's clearly intentional, but it's crazy to take this position on PBO that is a single toggle offered up on the first page of the BIOS by AMD themselves.
Associate
PBO over volts the CPU to 1.5v and over sometimes, its quite aggressive with CPU voltage, looks good because some get 5ghz plus on single core but it wont be doing your CPU much good.
Been busy building up my new 5600x and Crosshair Impact this weekend, so apologies for the lack of encouragement from me for the last couple of days. Its built now and initial power on tests are good. Very good...
4133mhz / 2066FCLK with G.Skill 4000mhz 16-19-19 2 x 16gb kit @ XMP. This is not properly stable, I have whea errors but it was at default volts so honestly this is way better than I expected. This is also not the limit, but I thought I'd grab a quick screenie to post up to show some progress. This also backs up my argument for the GamersNexus testing being terrible, the memtest time in ryzen dram calc is a full 26 seconds (over 20%) slower than my tweaked 2 x 8gb single rank Patriot Vipers @ 3733mhz... This cpu shows real promise though, spoiler - I'm posting this reply at 4266mhz mem / 2133FCLK with only a 50mv increase in vDIMM, vSOC and vIOD. I should be able to get some BIG memory numbers out of this cpu. I need to update all my test apps, pretty much nothing recognises Ryzen 5000 on my install
4133mhz / 2066FCLK with G.Skill 4000mhz 16-19-19 2 x 16gb kit @ XMP. This is not properly stable, I have whea errors but it was at default volts so honestly this is way better than I expected. This is also not the limit, but I thought I'd grab a quick screenie to post up to show some progress. This also backs up my argument for the GamersNexus testing being terrible, the memtest time in ryzen dram calc is a full 26 seconds (over 20%) slower than my tweaked 2 x 8gb single rank Patriot Vipers @ 3733mhz... This cpu shows real promise though, spoiler - I'm posting this reply at 4266mhz mem / 2133FCLK with only a 50mv increase in vDIMM, vSOC and vIOD. I should be able to get some BIG memory numbers out of this cpu. I need to update all my test apps, pretty much nothing recognises Ryzen 5000 on my install
Associate
