• 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.

5950x optimisation advice

TheSoupladel said:
I had two corsair AIOs fail on my which is why i ended up n with the Noctua, if i ever went back to water it would have to be custom loop so i dont have to repace the whole thing if a pump fails etc...
Click to expand...

The Corsair stuff is overpriced and isn't the best on the market (though rads and fittings are rebranded from other top manufacturers (Bitspower/HWlabs)) but it was my first custom loop and I got a good deal on a bundle price.

Happy enough and it does the job. Now, ICUE, that's a whole different topic. Version 3.x just about works, but version 4.x is the worst software launch I think I've ever seen. Literally broken, buggy and not even fitting to be called beta.

And Corsair thought it was a good idea to stop supporting a load of their own products with 4.x simply because they're a few years old.

The last issue doesn't affect me, but I quickly rolled back to 3.x anyway as version 4.x is unusable.
 
i am currently running with an RTX3060 but my hope is before the year is out and things perhaps get better i might end up with something better as i have a QHD monitor incoming tomorrow, in that event i might investigating putting in a custom loop to include the GPU in it.
 
TheSoupladel said:
The fault in my 5950x has been confirmed and i have a new chip incoming, hopefully tomorrow. Fingers crossed this one is ok :)
Click to expand...

Damn, "nice one". I say that in inverted commas because it's not nice to get a faulty chip but it's also worthwhile you did RMA that. At first I was just thinking it was dodgy BIOS settings or maybe CoreCycler playing up.

I wonder if it's poor-binning that should never have made it into retail. Aka, not binned to meet retail specs but slipped through.

Makes you wonder how many other chips there are out there failing on default settings but maybe someone doesn't know because the BSOD/reboots are few and far between. Or maybe not even happening yet. Sometimes instability can take months to pop up during normal use.

I guess this is why I'd always recommend someone runs some stability tests even on default settings. Just to make sure. Especially when we're paying the kind of money we are for something like a 5950x.
 
i have seen a couple of videos on YouTube which talk about anecdotal evidence of higher than normal failure rates in the 5000 series, so my hope is that the replacement will be ok.

I also hope that you are right and the bad chip was perhaps poor binning.

because i have had to go back to my 3900x, i decided to run corecycler on it and sure enough it is flawless which just reinforces my confidence that that everything else in my system is as it should be
 
TheSoupladel said:
i have seen a couple of videos on YouTube which talk about anecdotal evidence of higher than normal failure rates in the 5000 series, so my hope is that the replacement will be ok.

I also hope that you are right and the bad chip was perhaps poor binning.

because i have had to go back to my 3900x, i decided to run corecycler on it and sure enough it is flawless which just reinforces my confidence that that everything else in my system is as it should be
Click to expand...

Spot on, good idea running your 3900x through Corecycler.
 
The press are talking about CPU's being DoA. So that is not normal.

https://www.pcworld.com/article/3608349/ryzen-5000-failure-rates-we-reality-check-the-claims.html#:~:text=Ryzen 5000 series fails at 2.9 percent. Ryzen,For comparison, the company' data on Intel chips:

The kerfuffle kicked off on Sunday, when system builder PowerGPU Tweeted to its 222,000 followers: “The failure on the new AMD CPUs are still too high.” The company said that of the 320 Ryzen 5000-series CPUs it had received, 19 were “DOA” (dead on arrival), an eyebrow-raising six-percent failure rate. The company also said it had seen three to five failing B550- and X570-based motherboards a week.
Click to expand...

This next part,

The company said it isn’t seeing PowerGPU’s reported failure rates with its own systems. Interestingly, however, the vendor actually shared data indicating that Ryzen parts are failing the company’s internal quality screening at a higher rate compared to Intel chips—almost three times as high:
  • Ryzen 5000 series fails at 2.9 percent.
  • Ryzen 3000 series fails at 3 percent.
  • ThreadRipper 3000 series fails at 2.5 percent.
For comparison, the company' data on Intel chips:
  • Intel 9th-gen fails at 0.9 percent.
  • Intel 10th-gen fails at 1.2 percent.
Click to expand...

Not everyone gets problems.

Overclock3D.net’s Mark Campbell spoke with a UK PC builder who said it had seen no dead CPUs. “Our source claimed that they had experienced no DOA CPUs for the Ryzen 5000 series, a claim that either makes PowerGPU incredibly unlucky or our source incredibly fortunate,” Overclock3D.net said.

said it spoke with a popular local retailer. “...I’ve been told the failure rate for each part is normal and under 2%...”

One last report comes from Kitiguru.net’s Joao Silva, who cited RMAs from European store Mindfactory. The Ryzen 9 5950X rate was 0.77 percent; Ryzen 9 5900X, under 0.37 percent; Ryzen 7 5800X, 0.58 percent ; and Ryzen 5 5600X, 0.5 percent.
Click to expand...
 
zx128k said:
The press are talking about CPU's being DoA. So that is not normal.

https://www.pcworld.com/article/3608349/ryzen-5000-failure-rates-we-reality-check-the-claims.html#:~:text=Ryzen 5000 series fails at 2.9 percent. Ryzen,For comparison, the company' data on Intel chips:



This next part,



Not everyone gets problems.
Click to expand...

I think that's one of the articles i read

the new chip has arrived and has been installed and apart from switching on DOCP, its currently at stock and am putting it through an initial cycle of Corecycler
 
Well, i have discovered that because my motherboard has PBO settings in two places, i need to enter the values in both areas, i had PBO disabled in the extreme tweaker menu and was doing all my work in the AMD overclocking menu but i kept getting hard crashes and the TDC limit hitting 100% despite being set at 200. so entered the same values in both areas and reset the curve optimiser to -10 all cores and doing a run on corecycler, seems to be going ok now and no hard crashes or errors
 
TheSoupladel said:
Well, i have discovered that because my motherboard has PBO settings in two places, i need to enter the values in both areas, i had PBO disabled in the extreme tweaker menu and was doing all my work in the AMD overclocking menu but i kept getting hard crashes and the TDC limit hitting 100% despite being set at 200. so entered the same values in both areas and reset the curve optimiser to -10 all cores and doing a run on corecycler, seems to be going ok now and no hard crashes or errors
Click to expand...

Under the AMD Overclocking menu TDC and EDC have a hard cap. Its 160 TDC and 190 EDC.
 
well i have now got to a stage i am happy with, for now, and have cores that are boosting to 5.125Ghz!!

I am getting 4.475 under sustained multicore loads i.e Cinebench r23 but only getting a score of just shy of 29000 where as i know others have had 30k or over, now i know this is a synthetic benchmark but it maybe suggest there is more headroom to squeeze some extra performance out of this chip. My undervolts in curve optimiser aren't as high as i would like but i have gotten to the stage of -5 on one core, -10 on three cores and -15 on the rest and it be completely stables so i will leave at that for just now until i feel the urge to start fiddling with it again.
 
TheSoupladel said:
well i have now got to a stage i am happy with, for now, and have cores that are boosting to 5.125Ghz!!

I am getting 4.475 under sustained multicore loads i.e Cinebench r23 but only getting a score of just shy of 29000 where as i know others have had 30k or over, now i know this is a synthetic benchmark but it maybe suggest there is more headroom to squeeze some extra performance out of this chip. My undervolts in curve optimiser aren't as high as i would like but i have gotten to the stage of -5 on one core, -10 on three cores and -15 on the rest and it be completely stables so i will leave at that for just now until i feel the urge to start fiddling with it again.
Click to expand...

Not too dissimilar to me. One of my best cores needs -5. I've got -30 on the weaker cores, so -15 on the rest for you might be able to go higher.

What PPT, TDC and EDC did you end up using?
 
PPT 260, TDC 210 and EDC 190

I can probably drop the PPT and TDC a little because when running a multicore test in r23, i am hitting 90% and 76% respectively on those, but it is running quite toasty when at full load like that but never quite hits 90. At idle its sitting around 35 and under gaming load its usually between 60-70 but my room/office his quite a high ambient temperature right now because this house is too damn well insulated and the heat has nowhere to go, lol.
 
one oddity i just discovered was i tried turning on fmax enhancer and run r23 for a multicore test and sure enough, the speeds were much higher at around 4.65 but the overall score was worse, does that have anything to do with EDC dropping off the grid as a result of fmax enhancer being enabled?
 
TheSoupladel said:
one oddity i just discovered was i tried turning on fmax enhancer and run r23 for a multicore test and sure enough, the speeds were much higher at around 4.65 but the overall score was worse, does that have anything to do with EDC dropping off the grid as a result of fmax enhancer being enabled?
Click to expand...

Don't use Fmax enhancer, it's a 3xxx chip feature.

You don't need to hit 90~100% on PPT and TDC, so your TDC can likely come down, not hurt clocks and save some temps. PPT matters less.
 
You must log in or register to reply here.
Back
Top Bottom