Zen 2 for Distributed Computing: Any interest?

[Markfw]

You've lost me Tony! Must be a US thing

Btw, I've finally pulled the Trigger, I bought the MSI B450 Gaming Pro Carbon AC , 32GB Ballistix 3200 & a Ryzen 5 3600 , £453 from Ebuyer & Aria (CPU only).

Oh crap, I've just thought, will my PSU be compatible?? Err, I can't remember what it is, a Corsair 650w? Where was I talking about that recently?........

Just make sure you have an 8 pin and a 4 pin for the CPU. several times lately it has proved that you need both if the motherboard has both.

 

[biodoc]

Just make sure you have an 8 pin and a 4 pin for the CPU. several times lately it has proved that you need both if the motherboard has both.

That make sense if you are an enthusiast and are overclocking to the limits of the CPU. I run my CPUs at base clock so I only use the 8-pin connector and have had no problems.

Prompted by the news of a Ryzen 3900 (without -X) in early October, somebody tried different package power targets on a 3900X and reported Cinebench 20 results along with at-the-wall power consumption in a German forum. The test system had a lower-end HD 7750 GPU, a 650 W titanium-rated PSU, and was of course supplied from 230 V. Firmware was AGESA 1.0.0.3 ABBA.

The user calculated CB-points-per-Watt for package power in his post. I am showing CB-points-per-Watt for at-the-wall power instead.

comment​	TDP (W)​	PPT (W)​	TDC (A)​	EDC (A)​	CPU package power (W)​	Cinebench 20 points​	relative performance​	system power (W)​	relative system power​	points/W of system power​	relative efficiency
stock​	105​	142​	95​	140​	127.5​	7360​	1.00​	189​	1.00​	39​	1.00​
hard limit​	105​	105​	95​	140​	107.6​	7119​	0.97​	161​	0.85​	44​	1.14​
eco mode​	65​	88​	60​	90​	89.1​	6942​	0.94​	137​	0.72​	51​	1.30​
hard limit​	65​	65​	60​	90​	66.6​	6426​	0.87​	107​	0.57​	60​	1.54​
sweetspot​	​	58​	60​	90​	59.4​	6125​	0.83​	98​	0.52​	62​	1.60​

The performance degradation at lower power targets is rather moderate. Consequently, the lowest PPT tested is the optimum for performane-per-Watt not only when looking solely at CPU power consumption, but also when taking all of the platform power consumption overhead into account. I wonder how representative this image rendering workload may be for other computing scenarios.

I upgraded my 3900X bios to AGESA 1.0.0.4 B and decided to do a couple of quick tests with WCG under full load. I set the bios to optimized defaults and then tried changing cTDP and PPT to the following,

cTDP: 65 W / PPT: 65 W resulted in CPU freq of 3.55 GHz (underclock)
cTDP: 65 W / PPT: 88 W resulted in a CPU freq of 3.85 GHz (essentially base clock speed)

I'll do some more tests after the WCG birthday challenge.

 

[Kiska]

Just make sure you have an 8 pin and a 4 pin for the CPU. several times lately it has proved that you need both if the motherboard has both.

I'll leave this here:

"We have the extremely unnecessary extra 4 pin"

 

[Markfw]

Well, there are at least 3 confirmed reports of problems, that were verified to be fixed by using the 4 pin as well, as the 8 pin. I am not going to search now, just putting it out there. I already said "I told you so" and was proved right 3 times now.

 

[Kiska]

Well, there are at least 3 confirmed reports of problems, that were verified to be fixed by using the 4 pin as well, as the 8 pin. I am not going to search now, just putting it out there. I already said "I told you so" and was proved right 3 times now.

Cause according to spec, an 8 pin should deliver at least 384W of power.... So the extra 4 pin shouldn't even be required, unless the BIOS is doing something incredibly stupid.... Or its doing the MSI thing and instead of shoving a 6 pin power connector for PCIe power its grabbing it from the EPS rail

 

[Markfw]

Cause according to spec, an 8 pin should deliver at least 384W of power.... So the extra 4 pin shouldn't even be required, unless the BIOS is doing something incredibly stupid.... Or its doing the MSI thing and instead of shoving a 6 pin power connector for PCIe power its grabbing it from the EPS rail

I have no idea why just 8 pin would not work. I just know for a FACT sometimes it doesn't.

 

[Assimilator1]

Fair enough, I'm sure I posted about my PSU recently, might have even been in this thread! Lol, anyway, I'll re-find out what it is....

 

[Markfw]

Fair enough, I'm sure I posted about my PSU recently, might have even been in this thread! Lol, anyway, I'll re-find out what it is....

BTW, not saying ALL motherboards must have the 4+8 pin, just that I know that SOME have a problem without both.

 

[Assimilator1]

Ok.
Finally found where I'd posted about my PSU, thought I had a PSU problem during the SETI Wow event, hmm, it appears I never really followed it up! Although I thought I had checked the plugs......
Anyway, it's a Corsair TX650w. This old review says it has 8+4 pin plugs, as does an old CCL advert, I think!
So it seems it does, I think..... [update] No it doesn't, but it all works fine without that 4 pin socket being used!

 

[StefanR5R]

About the 3950X reviews which are gradually being published now: A few different reviewers will end up testing the very same processor specimen. For example, less than 5 reviewer kits have been sent to Germany (source).

 

[Howdy]

About the 3950X reviews which are gradually being published now: A few different reviewers will end up testing the very same processor specimen. For example, less than 5 reviewer kits have been sent to Germany (source).

I think when they are finally released, everyone will have a better chance at finding Unicorn poop or the pot of gold at the end of the rainbow. 3900x is still a issue to secure around me- Microcenter will show 10 for about an hour and then poof they are gone. They'd be gone before I drove to get one!

 

[crashtech]

I think when they are finally released, everyone will have a better chance at finding Unicorn poop or the pot of gold at the end of the rainbow. 3900x is still a issue to secure around me- Microcenter will show 10 for about an hour and then poof they are gone. They'd be gone before I drove to get one!

Well, camp out in front of the store then!

 

[Markfw]

I think when they are finally released, everyone will have a better chance at finding Unicorn poop or the pot of gold at the end of the rainbow. 3900x is still a issue to secure around me- Microcenter will show 10 for about an hour and then poof they are gone. They'd be gone before I drove to get one!

Newegg has them, and they are only $530. Amazon is like $550 !

 

[TennesseeTony]

Any pre-sales for the 3950X?

 

[Howdy]

Newegg has them, and they are only $530. Amazon is like $550 !

Micro center has them for $499

 

[Markfw]

Micro center has them for $499

Well, I know, but if you really want one, $30 is not bad.....

 

[crashtech]

Man, that 3900x is so tempting. But I'm stuck with what I've got until after the New Year.

 

[Howdy]

Well, I know, but if you really want one, $30 is not bad.....

I can wait for $30, I don't "need" another machine. I would just like to build another one. I can hold off since I want to build a complete machine. That money could buy more fans or something.

 

[StefanR5R]

less than 5 reviewer kits have been sent to Germany (source)

More precisely, 3 3950X samples and 2 TR3 samples. (source in German)

 

[StefanR5R]

Here are a few HPC oriented benchmarks with 3950X, stock and in eco mode:
review by former Tom's HW Germany

 

[Markfw]

Here are a few HPC oriented benchmarks with 3950X, stock and in eco mode:
review by former Tom's HW Germany

Well, I can't read it, its in German,m but I can see that even the 9980XE gets trounced most of the time.

 

[TennesseeTony]

Try opening the link using Chrome Browser, then select 'Translate' in the upper right corner.

 

[Markfw]

OK, I love this quote:

This workstation test shows very impressively that with the current Ryzen CPUs in the formerly lost positions (3D workspace, real-time preview) at least the connection could be restored and in all the areas where you have always been good (raytracing, Computing), even more points and could even continue to settle. AM4 vs. X299 is 1-0 today, who would have thought that a year ago?

 

[crashtech]

What's really wild is how they are managing to feed 16 cores with 2 memory channels!

 

[Markfw]

What's really wild is how they are managing to feed 16 cores with 2 memory channels!

I am very close to drooling. In 2 days (If I can pick my memory up) I will have 8 glorious channels of ram, and 128 threads in one box ! And the box is capable of 256 threads when I get the money !