-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
FRONTPAGE AMD Ryzen 9 3900X and Ryzen 7 3700X CPU Review
- Thread starter Overclockers.com
- Start date
- Joined
- Nov 21, 2018
I'm still trying to figure out why I have such dramatically worse temperatures than you do...
Can you humor me and run the latest AIDA64 stress test (as is) and report temps/post a screenshot after 30 mins, mack?
I got tons im trying to figure out.
Using 2 diff bios now. Changed nothing. Looked atr my voltages like i do every day. Now cpuz is reporting the same are core temp. What is high 1.4's using auto with pbo and a -.1v offset. Hwifo is close. Showing around .07v lower then everything else. Looks like i gotta bring my fluke home from work and check it.
And a manual oc 1.23v set in bios now shows in cpuz and coretemp 1.1v. So I know my CPU is not pulling 4.3ghz @ 1.23v.
I don't mind the innaccurate readings but man you gotta be on top of this. I think it's only happening to me. But I reboot a few times nothing. Made my system bsod reboot cpuz now reading 1.236v core temp is still saying "Bro it's totally 1.1v" hwmon 1.1v hwinfo 1.1v adia64 1.248.
Ok back to how it was. I don't get why it does this. I been pushing for 4.8 on all cores. I don't think windows is happy with the bclks I been running. I might have some messed up files lol. Who knows. I'm about to just go back to a 4.45 daily oc just so I know my voltage is locked and I'm sure what it is
Last edited:
- Joined
- Dec 18, 2000
We do... but this confirms it with even more samples with more empirical testing. Just adding more information, Dave!
It's just very disappointing to me as I don't enjoy building a system if I can't mess around with it. And I don't have any interest in playing with RAM timings since so little is gained performance wise. Unless I can find a real good deal on one of these, I might just pass on Zen 2 altogether.
- Joined
- Nov 21, 2018
Im very disappointed with my friend from Italy. Not only is he disabling SMT but now hes only clocking up a few cores. Oh well in 24 more hours i'll be putting up a 4.725 for him.
I'll say his 3900x is good.
I really need to work on my memory. I think im going to start that tomorrow. I think im gonna pull out 2 sticks and just run that around 3600. I really dont need 4 sticks of memory. I only play world of warcraft. And going back to wow classic woohoo 2006 here i come.
I'll say his 3900x is good.
I really need to work on my memory. I think im going to start that tomorrow. I think im gonna pull out 2 sticks and just run that around 3600. I really dont need 4 sticks of memory. I only play world of warcraft. And going back to wow classic woohoo 2006 here i come.
- Joined
- Jul 31, 2005
This isn't a stock PBO but an all core/smt 41x with ~1.225v fixed vcore in BIOS with a lower LLC setting. AIDA stress with just CPU/FPU. Cooling is custom water with a thick 360 rad... Ambient ~72F.
- Joined
- Mar 7, 2008
I'm still trying to figure out why I have such dramatically worse temperatures than you do...
Can you humor me and run the latest AIDA64 stress test (as is) and report temps/post a screenshot after 30 mins, mack?
I'm about to go away until the weekend so it'll be then. Also can you let me know what should/shouldn't be checked in the stress settings as I've likely changed it from default previously.
- Joined
- Aug 14, 2014
Trying for 4.8ghz on all cores and 4.45ghz daily on a Ryzen ? Prime95 stable ? On an AIO ? That must be some monster golden chip
- Joined
- Jul 15, 2006
So I was thinking (yeah, I know, bad sign) about Ryzen 3 and EPYC 2 and what that means for binning. The way things used to work is that there were different die designs for desktop and server CPUs. I mean the i7/i9 are not using the same die as the Xeons, even when they are based on the same CPU architecture.
This is something that has changed with Zen 2. The IO dies are different between Ryzen 3 and EPYC 2, but they use the exact same chiplets (core complex dies). Now if you look at what AMD has in the EPYC SKU list there are some sweet CPUs there, like the EPYC 7742. This thing is 64 cores at 2.25 GHz with a TDP of 225 W. That's like 3.5 W per core, not accounting for the IO die. That's gotta be some pretty sweet silicon. One might even call it epic.
Since EPYC 2 and Ryzen 3 use the same chiplets, this means they are competing with each other for the same binned dies. And given how much larger the margins are on those EPYC CPUs (even on a per chiplet basis) it makes sense that all the highly binned stuff would go to EPYC. That would mean Ryzen would be stuck with the runts of the litter.
This means that there might very well be some really sweet golden chips for Zen 2 out there, but that they are all going into EPYC. So our only chance of ever seeing them in Ryzen is if AMD somehow screws up their binning process.
This is something that has changed with Zen 2. The IO dies are different between Ryzen 3 and EPYC 2, but they use the exact same chiplets (core complex dies). Now if you look at what AMD has in the EPYC SKU list there are some sweet CPUs there, like the EPYC 7742. This thing is 64 cores at 2.25 GHz with a TDP of 225 W. That's like 3.5 W per core, not accounting for the IO die. That's gotta be some pretty sweet silicon. One might even call it epic.
Since EPYC 2 and Ryzen 3 use the same chiplets, this means they are competing with each other for the same binned dies. And given how much larger the margins are on those EPYC CPUs (even on a per chiplet basis) it makes sense that all the highly binned stuff would go to EPYC. That would mean Ryzen would be stuck with the runts of the litter.
This means that there might very well be some really sweet golden chips for Zen 2 out there, but that they are all going into EPYC. So our only chance of ever seeing them in Ryzen is if AMD somehow screws up their binning process.
- Joined
- Nov 21, 2018
Trying for 4.8ghz on all cores and 4.45ghz daily on a Ryzen ? Prime95 stable ? On an AIO ? That must be some monster golden chip
Nah custom loop. 3 thick 240's
Highest 24 totally prime stable is 4.675 on all cores. 4.725 crashes prime after about 10 min on a few cores. 4.8 at the most would just be maybe stable enough to do a benchmark. Right now I'm just making a guy from Italy push his.
- Joined
- Dec 19, 2012
I have no idea why my better cooling using a 'less stressful test' in AIDA64 yields 30C more. This is across multiple boards, mind you. I can't say I like the mount for AM4 on the H150i, but right now I'm using the Intel 115x mount on the ASRock X570 ITX board and the temperatures are remarkably similar... so its not the mount it seems. TIM applications are consistent.
P95 Small FFT yields temps to the mid 90s.
EDIT: It honestly feels like the cooling on these is, like many complained about Intel, just 'good enough', but not great... it's weird.
We do... but this confirms it with even more samples with more empirical testing. Just adding more information, Dave!
When stress testing you have to watch the CPU speed and voltage. I still say HWInfo is the most accurate, compare it to Ryzen Master if in doubt because it IS accurate. Back to the weird results.
Example the board I'm working on now stock CPU settings:
AIDA64 ST 70° 206W 4050 MHz 1.3V
AIDA FPU 71° 230W 3950 MHz 1.25V
P95 SFFT 65° 220W 3800 MHz 1.125V
These variations are from Performance Boost which varies CPU speed according to load, this behavior also varies with BIOS version and the motherboard you are using as an example the GODLIKE runs at 3950 MHz during SFFT and the CHVIII runs at 3900 MHz making comparing "stock" CPU behaviour impossible to test across motherboards.
- Joined
- Jul 31, 2005
I've been wondering about the binning process as well...
Do the EPYC chips run with any kind of boost?... I thought servers just run at default specified stock speeds.
With heat being trapped in smaller die shrinks maybe binning now favors higher leakage (cooler) running silicon more so than low leak (hotter) samples? Or do the low leak samples run at a specified default speed with less power consumed than higher leakage ones giving them a binning advantage?
The 3900X has a default P-State for 3.8GHz of: [FID 0x898 - VID 0x48 (38.00x - 1.100 V)]
Another thing I was curious to know was if the chiplets are all eight core or are there both six core and eight core chiplets being produced/fabbed?
*EDIT*: I found EPYC on Newegg and it shows a max boost for the 7742 chip @3.4GHz... I wonder how high a typical all core/smt boost (PBO) speed gets up to for these server systems?
https://www.newegg.com/p/N82E16819113581
Last edited:
On my GB and ASRock boards, hwinfo64 is showing results below 100W under load, similar on R5 3600 and R7 3700X. I know it has to be much more so I wouldn't trust these readings.
Silicon Lottery has posted their results so far of Ryzen 3 3700x/3800x/3900x results.
These things are all absolute duds when overclocking.
It's funny, after all the talk about a golden chip, bigazn's is the golden sample of golden samples if 6% only reach 4.2 GHz, LOL!
View attachment 206938
The 3700X I have does 4.25 GHz at 1.33V and is topped out thermally with both a MEH 2x120mm (EVGA 240 CLC) and a good 3x120mm (Corsair H150i)...as others have mentioned, such little difference, even with good cooling.
The best chips which caseking.de is selling under der8auer brand can make 4.3GHz on 3900X/3800X/3700X and 4.2GHz on R5 3600/3600X so not much better than the silicon lottery. My 3700X can make 4.3GHz 1.37V but 4.4GHz+ won't work stable regardless of settings (280 AIO for cooling and temps below 90°C).
- Joined
- Dec 19, 2012
Those power readings were from the wall Woomack
- Joined
- Mar 7, 2008
Can you humor me and run the latest AIDA64 stress test (as is) and report temps/post a screenshot after 30 mins, mack?
Ok, so didn't do it on weekend, but just done it now. No idea what room temp is, assume around 25C or slightly below. This is 3600 with Noctua D9L, back on thermal paste not the graphite pad, in a NZXT Manta with two 140mm intakes, no exhaust except from the D9L which is kinda pointed at the rear vent.
If it's at stock settings then I get about 10°C higher temps on the same cooler, with the same CPU in similar conditions. After adding 2nd fan to the D9L I get about 8°C lower temps. Actually, I get the same temps on 3600 and 3700X, even though there are 2 cores difference.
- Joined
- Aug 16, 2009
BBINNING
I thought I would get to this sooner but …. well life goes on. Dropped in some new ram HypeX Predator 4000 running at 3600 Cl 14 right now.
So at 1.25V load with prime running for 30 min I could get to 4.25GHZ. Now I wouldn't call 30 min stable but it was all I can do tonight. When I went up to 4.275 Prime dropped worker 24 at 11min and another worker at 17min so it is not far from being marginally stable at 4.275. The issue is we don't know how exactly Silicone Lottery is binning. Take this for what it's worth.
I thought I would get to this sooner but …. well life goes on. Dropped in some new ram HypeX Predator 4000 running at 3600 Cl 14 right now.
So at 1.25V load with prime running for 30 min I could get to 4.25GHZ. Now I wouldn't call 30 min stable but it was all I can do tonight. When I went up to 4.275 Prime dropped worker 24 at 11min and another worker at 17min so it is not far from being marginally stable at 4.275. The issue is we don't know how exactly Silicone Lottery is binning. Take this for what it's worth.
Similar threads
