AMD Ryzen 5000 Builders Thread

[B-Riz]

	Cores Threads	Base Freq	Turbo Freq	L3 Cache	TDP	MSRP
Ryzen 9 5950X	16c/32t	3400	4900	64 MB	105 W	$799
Ryzen 9 5900X	12c/24t	3700	4800	64 MB	105 W	$549
Ryzen 7 5800X	8c/16t	3800	4700	32 MB	105 W	$449
Ryzen 5 5600X	6c/12t	3700	4600	32 MB	65 W	$299

5600X comes with cooler, the rest do not.

above from https://www.anandtech.com/show/1614...en-3-on-nov-5th-19-ipc-claims-best-gaming-cpu

 

[Hitman928]

Well yes, that's the prerequisite for it to be able to reach higher frequencies than 3800X.

Then that doesn't explain the much higher temps because the lowered voltage is off setting the higher frequency so that the power consumption is not much higher in the 5800x compared to the 3800x. There are also reviews out there with 5800x samples where they aren't reporting these very high temps.

 

[shinjuku]

i'm about to buy the b450 tomahawk max motherboard (this seems to be the best value option for me) to go with my 3600, and just wait for the bios update for the 5000 series. I plan to get the 5600x or 5800x whenever it comes back in stock though. my goal is to play PC games/video editing/photo editing raw files as a hobby.

i don't know much about PC components and performance. will getting a b450 to go with 5600/5800x affect performance (lower fps in games, lower video rendering times, etc) as opposed to getting a b550 w/ 5600x 5800x? or will it be the same. thanks.

 

[Makaveli]

If the price difference is small I would go B550 over B450.

 

[moinmoin]

Then that doesn't explain the much higher temps because the lowered voltage is off setting the higher frequency so that the power consumption is not much higher in the 5800x compared to the 3800x. There are also reviews out there with 5800x samples where they aren't reporting these very high temps.

The problem is not (package) temp but hotspots, not unlike experienced on Zen 2, just that the denser logic and higher frequencies on Zen 3 makes them appear more often/stronger. And among the four Zen 3 based chips released so far 3800X is the one attracting them the most due to the high TDP on just one chiplet.

 

[Hitman928]

The problem is not (package) temp but hotspots, not unlike experienced on Zen 2, just that the denser logic and higher frequencies on Zen 3 makes them appear more often/stronger. And among the four Zen 3 based chips released so far 3800X is the one attracting them the most due to the high TDP on just one chiplet.

Sure, and that would explain an increase in temp from say a 5950x to a 5800x at stock, but what is hard to accept is a 15 - 20 degree temp increase from 10% higher clocks, even with water. As an example, OC3d overclocked the 5950x to 4.7 GHz (higher than a 5800x at stock) and with over 200 watts being generated it rises to 'only' 86 C on a CLC. I have a very hard time believing a 5800x should be hitting 90C on an open loop at stock.

 

[Hitman928]

@MrTeal Can you report voltage and frequency at 100% under y-cruncher when the 5800x is set to stock?

 

[phillyman36]

Sitting here with everything but the 5900x and mobo. If i did catch covid 19 i would have the 5900x from Microcenter. sad face.

 

[MrTeal]

@MrTeal Can you report voltage and frequency at 100% under y-cruncher when the 5800x is set to stock?

Bone stock with XMP off, or no-PBO but XMP enabled?

 

[Hitman928]

Bone stock with XMP off, or no-PBO but XMP enabled?

XMP off then on would be great, no PBO either time.

 

[MrTeal]

Sure, I can run that this evening.

 

[Justinus]

I just managed to snag a 5600x on Amazon.... by spamming the "Buy Now" button from the in-stock version of the page that was locally cached on my phone after the product was shown as out of stock.

I used this knowledge to create a local override of the 5950x page that exposes the Buy Now button, and when amazon starts making changes to the 5950x page, indicating they are beginning to stock them, I'm going to spam the "Buy Now" button until it takes.

If it works.
In order to beat the bots, you must think like the bots.

Disclaimer: I fully intend to either cancel my 5600x order if I snag a 5950x or use it as a holdover CPU and sell it for a reasonable price. No scalping or greed here, I just need to get this box up and running asap.

A minor update: I learned under fire (5950x restock) my method does not work unless I have access to the Offer Listing ID for the item. Can't acquire it without having the source from the in-stock version of the item page. So I guess getting the 5600x was luck and a one-off, unless there is ever stock of the 5950x long enough for the page to update to "in stock".

 

[lightmanek]

So I've spent few hours playing with my new processor and tweaking it for Single Threaded performance using PBO.
What shocked me was that my mobo allows to set +300MHz above what stock CPU boosts to, resulting in 5175MHz single core boost. I have at least two cores which consistently boost to 5150MHz now, load dependent.

This setting is not the best for MT, as PBO gives a bit too much voltage, but it still is under 190W CPU package power.

Pay attention to Cinebench OpenGL score rocketing by 30FPS after slight tweak to memory and ST core clock
Also CPU-Z jumped by over 30 points compared to stock, my memory latency has dropped to 56ns as well. All that without even going into FSB overclocking - something to play with tomorrow

 

[Justinus]

So I've spent few hours playing with my new processor and tweaking it for Single Threaded performance using PBO.
What shocked me was that my mobo allows to set +300MHz above what stock CPU boosts to, resulting in 5175MHz single core boost. I have at least two cores which consistently boost to 5150MHz now, load dependent.

This setting is not the best for MT, as PBO gives a bit too much voltage, but it still is under 190W CPU package power.

Pay attention to Cinebench OpenGL score rocketing by 30FPS after slight tweak to memory and ST core clock
Also CPU-Z jumped by over 30 points compared to stock, my memory latency has dropped to 56ns as well. All that without even going into FSB overclocking - something to play with tomorrow

Can you grab the new Cinebench R23 and run it for us? Feel free to enable advanced mode and kill the timer. I don't care about 10 minutes of runs, just the initial score for both ST and MT.

 

[Noid]

Can you grab the new Cinebench R23 and run it for us? Feel free to enable advanced mode and kill the timer. I don't care about 10 minutes of runs, just the initial score for both ST and MT.

During installer execution, required an Email, and accepting EULA crap.
This is what you see after you validate your Email.

So, they have successfully acquired my Email, and installed nothing.
Beware .... removed this spyware ...
{edit} I got mislead at the direct D/L link for the Cinema 4D R23.110 Auto-Installer (PC)
However, it still did a bait - n - switch on me.
Does this mean the benchmark is not doing the same ? hmmm

 

[lightmanek]

During installer execution, required an Email, and accepting EULA crap.
This is what you see after you validate your Email.

So, they have successfully acquired my Email, and installed nothing.
Beware .... removed this spyware ...

I hope you are aware that you just installed FULL VERSION of professional 3D Animation suite from Maxon and not their CineBench benchmark tool?

 

[lightmanek]

Can you grab the new Cinebench R23 and run it for us? Feel free to enable advanced mode and kill the timer. I don't care about 10 minutes of runs, just the initial score for both ST and MT.

I had higer ST in R23 of 1608 points but had to clear BIOS when playing with FSB (101.5 a no go with my PCIe 4.0 NVME) so I've not yet set my 2nd memory timings as they were.
Anyway a quick run for you.

 

[Noid]

Hey, did you notice the multipliers in your new AIDA64 ?
( 51.4 and 46.5 ---- this new version bugs me )
Nice FCLK !

 

[lightmanek]

Hey, did you notice the multipliers in your new AIDA64 ?
( 51.4 and 46.5 ---- this new version bugs me )
Nice FCLK !

Yes, normal behaviour. One is reporting max. multi during multicore test and the other my maximum single core multiplier.

With upcoming BIOS, if I can get to 2000MHz IF and tweak for higher multicore clocks using PBO, results will be even better.
Remember, all my tests are on 1 year old installation of W10 with lots of services running in the background. Can't be bothered to remove them as they needed.

 

[Noid]

2000

{edit} dumb Q removed - I see the minimum duh

Only one x570 result listed in AIDA64 " CPU Queen " as comparison.
Will be interesting to see more numbers for the x570.

 

[lightmanek]

1.48v - just saw your voltage ... ! --- is that static ? or Auto ?

Only one x570 result listed in AIDA64 " CPU Queen " as comparison.
Will be interesting to see more numbers for the x570.

All above is achieved using PBO, so it's Auto and under load is more like 1.288V, only ST loads spike to 1.45V+

 

[CakeMonster]

So it looks like OC'ing the CPU is pretty pointless for gaming performance (GN and HU reviews showing no difference, even reduction). Would one of the cheaper ASUS X570 boards still allow for tweaking the last bit out of PC3600 RAM?

 

[Noid]

Doh

 

[Noid]

I forgot the base clock on the 5900 is 3.7

 

[lightmanek]

So it looks like OC'ing the CPU is pretty pointless for gaming performance (GN and HU reviews showing no difference, even reduction). Would one of the cheaper ASUS X570 boards still allow for tweaking the last bit out of PC3600 RAM?

Not exactly. If you want static all core OC, then yes, in most cases you will have to sacrifice ST clock and therefore gaming performance. There are few exceptions and in case of 5600X some samples can clock all cores to max ST boost or even above.
In other cases use PBO same as I'm doing to gain performance that way. I'm getting +200MHz across all loads on average at the expense of power usage. Still, tweaking memory and overclocking IF links brings biggest improvements for gaming though.

Check my OpenGL scores from Cinebench I've posted as this is representative of a lot of gaming improvements too (15%-20% improvement going from untweaked stock CPU with DOCP 3600 CL16 RAM to tweaked 3866 CL18, PBO, IF and FSB tweaks)

 

[Noid]

Sacrifice some nanos for more bandwidth. ( I'm hitting 57k - new AIDA )
See what the scores become ...