nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Flagged as Spam (1)
Has anyone found a fix for the 3090 not scoring as well on ryzen 5000? I just received a 5950x and msi meg ace x570 and it scores 600pts lower. Only thing i can figure is my oc on my 10900k is helping it out. 5200 core 4800 cache. The 5950x is nice just useless if i cant use it for benchmarking without losing score. Everything was equal minus 10mhz on mem. Same classfied settings and all. 1200w psu in the amd pc and 1600w psu in 10900k pc. Both have msi meg ace for respective chip sets. The Intel build did have the 64gb ram before. Dropped back to the 32gb. Maybe I need faster ram for the amd? But I know I'm not the only one that saw a performance hit. https://www.3dmark.com/co.are/pr/892316/pr/892171#
post edited by nova496 - 2021/02/19 08:35:34
|
Xinoxide
New Member
- Total Posts : 27
- Reward points : 0
- Joined: 2020/10/09 15:25:11
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 06:40:16
(permalink)
This is the most pointless.
Have a good day guy.
If you must compare than you just have to have and maintain both. Only way you'll feel good Inside.
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 08:32:46
(permalink)
Xinoxide
This is the most pointless.
Have a good day guy.
If you must compare than you just have to have and maintain both. Only way you'll feel good Inside.
How is it pointless? I am curious what made my 3090kpe lose score. Is it pointless because you don't have a kpe? Or a 5950x ryzen? I'm trying to see if anyone had a fix for a known issue. If you think this is pointless then obviously you didn't think your comment through. As you offered nothing but to say my post was pointless? I'm not comparing them as you say. I legit want to know why other then cpu clock speeds if that's the only thing that is making my kingpin lose score. As its not just a few pts its 600pts. So if you must accuse it of being pointless keep on keeping on and don't bother replying.
|
arestavo
CLASSIFIED ULTRA Member
- Total Posts : 6916
- Reward points : 0
- Joined: 2008/02/06 06:58:57
- Location: Through the Scary Door
- Status: offline
- Ribbons : 76
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 08:42:36
(permalink)
From what I've read, there's bit of tweaking needed to get the most out of an AMD system. Disabling SMT should help. There's also something about matching the RAM speed to the CPU... I forget exactly what it is, some sub-frequency that runs at 1/2 of the RAM speed and can have a fairly large impact if it's not matched (or if the RAM is slow). Edit: also, using the same driver version should help eliminate that variable. You used two different versions in the comparison.
post edited by arestavo - 2021/02/19 08:45:47
|
Mathieas
SSC Member
- Total Posts : 585
- Reward points : 0
- Joined: 2009/03/28 23:11:56
- Status: offline
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 08:53:40
(permalink)
arestavo
From what I've read, there's bit of tweaking needed to get the most out of an AMD system.
Disabling SMT should help. There's also something about matching the RAM speed to the CPU... I forget exactly what it is, some sub-frequency that runs at 1/2 of the RAM speed and can have a fairly large impact if it's not matched (or if the RAM is slow).
Edit: also, using the same driver version should help eliminate that variable. You used two different versions in the comparison.
I believe you are thinking of the infinity fabric clock. Ram freqs up to 3600mhz with the correct matched ram will yeild the highest performance. After that you start to get unfavorable ratios that cause the infinity fabric clock to need to be dropped. This in general, can lead to lower over all performance even at higher ram speeds. I am not an expert OP and full disclosure I have never owned an AMD cpu. I was considering it and thus did some research. At the very least the above should at least give you some good keywords to search for and dig deeper :).
post edited by Mathieas - 2021/02/19 08:56:46
CPU: AMD Ryzen 5 5800x3D - Fclk@1800mhz Cooling: EKWB 360 AIOMother Board: X570 Asus Crosshair VIII FormulaGPU: 1x 4090 MSI SUPRIM Liquid XRAM: 2x 16GB G.Skill Trident Z Neo @3600mhz CL14 (B-Die) SSDs(NVME): 1X WD sn850x 4TB, SSDs(SATA): 1X Samsung 870 EVO 2TB, 1x Samsung 850 Pro 512GB, 1x Samsung 840 Pro 256GBSound: Creative Sound Blaster ZXR Case: Phantek P600SPSU: EVGA SuperNOVA 1000W T2Monitor(Main): Alienware AW3423DW (175hz QD-Oled)Monitor(2ndary): Asus VG259QR (165hz IPS) OS: Windows 10 pro 64bit
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 09:34:31
(permalink)
Yeah I thought about disabling SMT, and I know fabric clock with memory. My 3950x didn't lose performance like my 5950x did. Also that was after I ddu the driver it was the only driver I had available that showed highest score on my ftw3. https://www.3dmark.com/co.are/pr/892054/pr/892316#That is same driver on the same day.
|
Mathieas
SSC Member
- Total Posts : 585
- Reward points : 0
- Joined: 2009/03/28 23:11:56
- Status: offline
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 09:50:59
(permalink)
nova496
Yeah I thought about disabling SMT, and I know fabric clock with memory. My 3950x didn't lose performance like my 5950x did.
Also that was after I ddu the driver it was the only driver I had available that showed highest score on my ftw3.
https://www.3dmark.com/co.are/pr/892054/pr/892316#
That is same driver on the same day.
When looking at both of those i see a higher a boost clock and lower temps on the 10900k. Could this be a case heat/air flow issue? When you start PR do they both clock to the same value when the temp is lower?
CPU: AMD Ryzen 5 5800x3D - Fclk@1800mhz Cooling: EKWB 360 AIOMother Board: X570 Asus Crosshair VIII FormulaGPU: 1x 4090 MSI SUPRIM Liquid XRAM: 2x 16GB G.Skill Trident Z Neo @3600mhz CL14 (B-Die) SSDs(NVME): 1X WD sn850x 4TB, SSDs(SATA): 1X Samsung 870 EVO 2TB, 1x Samsung 850 Pro 512GB, 1x Samsung 840 Pro 256GBSound: Creative Sound Blaster ZXR Case: Phantek P600SPSU: EVGA SuperNOVA 1000W T2Monitor(Main): Alienware AW3423DW (175hz QD-Oled)Monitor(2ndary): Asus VG259QR (165hz IPS) OS: Windows 10 pro 64bit
|
xxDarkxx
Superclocked Member
- Total Posts : 135
- Reward points : 0
- Joined: 2020/11/02 07:51:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 10:15:03
(permalink)
yea i wont buy AMD until they show at least 15% better than intel equivalent I just run intel because in truth ram makes the intel cpu faster when i can run 4533 15 16 16 34 it just simply makes intel faster despite the media jumping all over amd, intel is still the best (save a few games here and there) there is no amount of tweaking that will make the 5950x perform better in port royal, just take a look at 3dmark top 100 it will have your answer
post edited by xxDarkxx - 2021/02/19 10:18:24
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 10:15:14
(permalink)
Mathieas
When looking at both of those i see a higher a boost clock and lower temps on the 10900k. Could this be a case heat/air flow issue? When you start PR do they both clock to the same value when the temp is lower?
The 10900k has 20pts at first turn on the 5950x. The first score there was a 1° difference second score was 2°. I can't get the 5950x system to hold the same boost clock. Everything is set the same just won't boost as high and hold. I'll buy some 3600mhz memory and see if that helps. I know the ryzen is finicky and can have instability issues higher then 3600mhz. I just wasn't expecting to lose 600pts.
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 10:16:29
(permalink)
I agree dark I bought it for testing purposes and a 20% lead in synthetic bench on cpu bound test, but falls short in gpu test.
|
Dabadger84
CLASSIFIED Member
- Total Posts : 3426
- Reward points : 0
- Joined: 2018/05/11 23:49:52
- Location: de_Overpass, USA
- Status: offline
- Ribbons : 10
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 10:32:14
(permalink)
You're also going to experience score fluxes like this with a similar card, just keep that in mind.
I.E. I've gotten as high as 15.4k on my card & setup, but right now I can't get higher than 14.9-15k at the exact same settings... on the same drivers... so it's either Windows gimping performance or just the luck of the draw on whether or not the system wants to perform "as well as it can" on a given boot up. A lot of other people have posted about this being the case in the Kingpin threads here on the forums.
I haven't been able to replicate a 15.4k score since the day I first got them, and I have no idea why. Even tried a fully fresh Windows install with just updates, drivers & 3DMark installed - still didn't get over the low 15.0k mark again.
ModRigs: https://www.modsrigs.com/detail.aspx?BuildID=42891 Specs:5950x @ 4.7GHz 1.3V - Asus Crosshair VIII Hero - eVGA 1200W P2 - 4x8GB G.Skill Trident Z Royal Silver @ 3800 CL14Gigabyte RTX 4090 Gaming OC w/ Core: 2850MHz @ 1000mV, Mem: +1500MHz - Samsung Odyssey G9 49" Super-Ultrawide 240Hz Monitor
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 10:41:37
(permalink)
Dabadger84
You're also going to experience score fluxes like this with a similar card, just keep that in mind.
I.E. I've gotten as high as 15.4k on my card & setup, but right now I can't get higher than 14.9-15k at the exact same settings... on the same drivers... so it's either Windows gimping performance or just the luck of the draw on whether or not the system wants to perform "as well as it can" on a given boot up. A lot of other people have posted about this being the case in the Kingpin threads here on the forums.
I haven't been able to replicate a 15.4k score since the day I first got them, and I have no idea why. Even tried a fully fresh Windows install with just updates, drivers & 3DMark installed - still didn't get over the low 15.0k mark again.
Legit did back to back tests lol. Card was still warm when pulled out of the amd system put back in my 10900k system and same settings wham 15291. I did a fresh install of windows ill try again see if it had any issues with it and retest. Could be a lot of variables. Memory frequency to low at 3200mhz or any other reason.
|
emmett
SSC Member
- Total Posts : 513
- Reward points : 0
- Joined: 2008/03/30 23:42:22
- Status: offline
- Ribbons : 2
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 11:06:17
(permalink)
I have same issue also. 9900k VS 5950X. I lose up to 300+ points using the 5950X.
On another forum someone stated that it is due to clocks and effective clocks. I can lock 5400 on 9900k but
not sure how to do that on ryzen if even possible (lock a single core at 5000 for single core). ryzen is very new to me.
If you look at HOF timespy scores, lots of ryzens there thanks to core count.
I have also been playing with riser cables, different lengths to see how they effect performance. and noticed that on the intel machine
if i add a second card of any kind to second slot, dropping both to PCIE 3.0 8X that also effects Porte royal score, it NEEDS PCIE 3.0 16X bandwidth.
In the ryzen machine PCIE 4.0 8X will net same score as PCIE 3.0 16x (pretty much equal bandwidth between the two.)
So if anyone is running PCIE 3.0 8X I believe your scores will suffer..
I am about to box up my 9900k system and wait for 11900k
Currently using Gigabyte Z370 Aorus Gaming 7 EVGA 2080 TI KingPin Intel 9900K GSkill 3200 EVGA G2 1300
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 11:16:33
(permalink)
emmett
I have same issue also. 9900k VS 5950X. I lose up to 300+ points using the 5950X.
On another forum someone stated that it is due to clocks and effective clocks. I can lock 5400 on 9900k but
not sure how to do that on ryzen if even possible (lock a single core at 5000 for single core). ryzen is very new to me.
If you look at HOF timespy scores, lots of ryzens there thanks to core count.
I have also been playing with riser cables, different lengths to see how they effect performance. and noticed that on the intel machine
if i add a second card of any kind to second slot, dropping both to PCIE 3.0 8X that also effects Porte royal score, it NEEDS PCIE 3.0 16X bandwidth.
In the ryzen machine PCIE 4.0 8X will net same score as PCIE 3.0 16x (pretty much equal bandwidth between the two.)
So if anyone is running PCIE 3.0 8X I believe your scores will suffer..
I am about to box up my 9900k system and wait for 11900k
Yeah pcie 3.0 x16 on intel and 4.0x16 on amd the multicore it's a beast. Port royal best I've seen on HOF is 15916. If I could break into low 15k I'd be happy. My intel will hit 15250-15300 all day repeatedly.
post edited by nova496 - 2021/02/19 16:38:59
|
professordumbdumb
New Member
- Total Posts : 30
- Reward points : 0
- Joined: 2020/05/23 16:41:55
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 16:22:08
(permalink)
This is an issue I also experience. Port Royal score scores are lower on amd.
Best run I had on a 2 pin asus tuf oc 3090 card was 146-- with a 9900k and very tightly controlled memory timings - I think avg clock was in the 2040-2060 range.
Now on a 5800x with a kpe card at 2160mhz I pull up at 14200 (though I've increased this recently with msvdd tweaking to above 15k).
Intel seems to have the lock on port royal - and I suspect it has something to do with memory latency (as this is the last major advantage intel has over amd). This is of course masked in gaming to a degree - and even a lot of synthetic benchmarks. Even a perfectly tuned amd system cannot touch a recent intel system tuned by a sausage. I've been playing with amd in a couple of systems - and unfortunately the two 5800x's I have exhibit terrible memory write performance - exactly half their read performance - as a consequence of the intentionally hamstrung memory controller when only one die is present on the package. 5900x and 5950x do not have this problem.
In any case - the best I can do with memory tuning on amd with 1:1:1 fclk / uclk / mem - is a latency of about 49ns in aida extreme running 2000mhz fckl with 14-14-14-28c1 4000mhz b-die with knife-edge sub timings, useful for a run or two but wholly pointless otherwise.
On intel, 49ns is like stock 3200mhz c16 ffs. I think the record is sub 24ns on hwbot.
Bottom line from a pointless rant is that I agree that amd seems to be down compared to intel in port royal with 3090. Will update when I can get rocketlake.
|
liud21
iCX Member
- Total Posts : 360
- Reward points : 0
- Joined: 2020/09/22 20:02:51
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 16:43:56
(permalink)
in other words, Intel is better than AMD's newest chip in benchmarks...
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 16:50:34
(permalink)
liud21
in other words, Intel is better than AMD's newest chip in benchmarks...
Honestly I was just seeing what it would do and since I have a kingpin and wanted to test if it stacked up against the 10900k it was replacing and fell short 600pts I wanted to see if there was anything that anyone found. Benchmarks don't show all the detail but there are people that have issues playing games and other things buddy had a 5800x and 3090ftw3 getting 20fps in cyberpunk and I was getting 60-80. Same monitor just different cpus. This isn't a passing match I was curious if anyone found work arounds.
|
kraade
CLASSIFIED Member
- Total Posts : 2369
- Reward points : 0
- Joined: 2020/10/18 09:05:52
- Location: Georgia
- Status: offline
- Ribbons : 7
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 17:07:57
(permalink)
That is why they are on SALE !
|
Roy10266
New Member
- Total Posts : 67
- Reward points : 0
- Joined: 2016/05/09 17:38:19
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 17:14:49
(permalink)
It's not the CPUs fault. 8600k vs 8700k I get the same score in Port Royal. It's a GPU benchmark.
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 17:37:54
(permalink)
Roy10266
It's not the CPUs fault. 8600k vs 8700k I get the same score in Port Royal. It's a GPU benchmark.
Did you click the link? 15291 on my intel build same kingpin card I swapped into my 5950x amd build that scored max 14720. I did the test on the amd build first took out the gpu still warm to the touch put back in 10900k build and score went back to 15k+. When I originally received my kpe I scored 14.2k with only with power target maxed. 13.6k with my amd build same settings.
|
Clovis559
SSC Member
- Total Posts : 563
- Reward points : 0
- Joined: 2020/09/18 13:53:08
- Status: offline
- Ribbons : 3
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 19:49:01
(permalink)
Keep cracking at it! Be interested to see if you fix it or if there is a difference. People in the top 100 with 5000 series, you think they'd switch CPU's if it got them that much higher. Dabadger84
You're also going to experience score fluxes like this with a similar card, just keep that in mind.
I.E. I've gotten as high as 15.4k on my card & setup, but right now I can't get higher than 14.9-15k at the exact same settings... on the same drivers... so it's either Windows gimping performance or just the luck of the draw on whether or not the system wants to perform "as well as it can" on a given boot up. A lot of other people have posted about this being the case in the Kingpin threads here on the forums.
I haven't been able to replicate a 15.4k score since the day I first got them, and I have no idea why. Even tried a fully fresh Windows install with just updates, drivers & 3DMark installed - still didn't get over the low 15.0k mark again.
I'm thinking of having a ssd just for benching 
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 20:18:09
(permalink)
Oh well did more testing, will have to order a kit of 3600mhz cas16. And figure out what to do. Maybe I'll just make my 10900k the benchmark rig and the amd the gaming streaming rig.
Thanks for the ideas. Tried smt off and overclocking just 2 cores nothing got me in the 15k. Have a good one and mods can close this thread
|
neteng101
Superclocked Member
- Total Posts : 153
- Reward points : 0
- Joined: 2018/02/03 06:56:57
- Status: offline
- Ribbons : 1
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 20:22:56
(permalink)
Thanks for confirming what I've thought all along - AMD still is bad if you really care about gaming performance. Its just so much easier to tune and overclock on Intel - the AGESA is still a mess and trying to get the 1:1:1 clock to work beyond 3800 good luck with that too. The 10900K was on sale earlier today around $420 - anyone thinking should have grabbed one. The Rocket Lake CPU is going to only be 8C/16T and not necessarily better, but Alder Lake CPU could be a total game changer.
Can't imagine why anyone with a 10900k would have attempted the switch to the 5950x, seems like a fool's gambit. All you have to do is look at the 3DMark scoreboard and you can see how the 5950X is getting soundly beaten by the 10900k in CPU scores all the time.
|
nova496
New Member
- Total Posts : 85
- Reward points : 0
- Joined: 2014/11/24 16:34:26
- Status: offline
- Ribbons : 0
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 21:12:05
(permalink)
neteng101
Thanks for confirming what I've thought all along - AMD still is bad if you really care about gaming performance. Its just so much easier to tune and overclock on Intel - the AGESA is still a mess and trying to get the 1:1:1 clock to work beyond 3800 good luck with that too. The 10900K was on sale earlier today around $420 - anyone thinking should have grabbed one. The Rocket Lake CPU is going to only be 8C/16T and not necessarily better, but Alder Lake CPU could be a total game changer.
Can't imagine why anyone with a 10900k would have attempted the switch to the 5950x, seems like a fool's gambit. All you have to do is look at the 3DMark scoreboard and you can see how the 5950X is getting soundly beaten by the 10900k in CPU scores all the time.
Honestly its funny how poor people think. How about I am a pc enthusiast, I have owned multiple Intel processors and have built a handful of AMD they have there place. What I see is I paid retail 799.99 for a cpu that is being scalped for way more. The best part is I didn't make this to bash AMD but to figure out somethings. I had no issues with any of my intel processors. the 10900k fit the bill when I replaced the the 9900k that died on me and I bought a 3950x. But the 10900k at 5200 alll 10 cores and 4800 cache getting crushed by 20% on my system when the amd is hitting 5ghz on 3 cores and 4.8 on the 13 ohter ones. I think its fair to say they have the there place. Does intel overclock ram easier yea does it over clock better yes is it a better CPU then amd in everything no. will the 11900k be that great probably not. Its like comparing my Phanteks Entho cools better then the lian li 011 dynamix xl with distro plate. I work for my money and try to not bash one thing over the other. I see it with the Job I am in. there is always going to be things that I see that don't make me think of buying other brands products but oh well man.
|
Sajin
EVGA Forum Moderator
- Total Posts : 49199
- Reward points : 0
- Joined: 2010/06/07 21:11:51
- Location: Texas, USA.
- Status: offline
- Ribbons : 199
Re: 5950x with 3090kpe vs 10900k 3090kpe
2021/02/19 21:14:35
(permalink)
Closing thread per op request.
|