Info 64MB V-Cache on 5XXX Zen3 Average +15% in Games

Page 150 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Kedas

Senior member
Dec 6, 2018
355
339
136
Well we know now how they will bridge the long wait to Zen4 on AM5 Q4 2022.
Production start for V-cache is end this year so too early for Zen4 so this is certainly coming to AM4.
+15% Lisa said is "like an entire architectural generation"
 
Last edited:
  • Like
Reactions: Tlh97 and Gideon

Dave3000

Golden Member
Jan 10, 2011
1,351
91
91
Got my 5800X3D the other day. I tried -30 on all cores and I can benchmark and stress test until the cows come home and no problem. The moment I load up 7 Days 2 Die, within the first few minutes the PC reboots. Go figure.

100% solid at -25 however, I'll take it.

This is the problem with overclocking or undervolting. It may seem stable for a week and frequent stress testing but a week later you may run into something that can expose instability of the overclocking or undervolting. There is nothing noticeable to be gained by overclocking CPU's manufactured in the past 5 years anyways and certainly not worth the efforts in my opinion.
 
  • Like
Reactions: CHADBOGA

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
Nobody have said that form factors will not change for future desktop computers.

The form factor is the whole reason CAMM exists. If they changed it, it would not be CAMM.

The whole purpose CAMM is a thing is to get away for the very limiting design that current RAM modules have where we end up with these really long traces.
 

MadRat

Lifer
Oct 14, 1999
11,910
238
106
Maybe they will take advantage of both sides of mainboards in future form factors. Seems like we are reaching the point where heat is pooling at the PCB. Could be interesting to see soldered in memory underneath the core and on the backside of the mainboard, to provide minimal trace lengths. CAMM to provide expansion memory on the topside of the mainboard, to provide minimal trace lengths there, too. Paired DDR5 memory channels could run to the CAMM, and the mainboard manufacturer can run memory channels under a flexible architecture to maintain sensitivity to market segmentation. A cheaper manufacture could forgo either the CAMM module or soldered in RAM altogether, offering just enough base memory with support for reduced memory channels under either architecture to run a system. (Maybe a POS equipment manufacturer finds one 32-bit memory channel desirable.) A mid-tier manufacturer may run four 32-bit channels to the CAMM module. Another mid-tier could run both, but go for two 32 bit DDR channels to each to keep it cheap. Another mid-tier may offer three 32 bit DDR channels to the soldered in memory, but only support a single 32 bit DDR channel for the CAMM module. Or they may just run conventional DDR5 through the four channels. Flexibility would be the key.
 
  • Like
Reactions: Vattila

Glo.

Diamond Member
Apr 25, 2015
5,705
4,549
136
The form factor is the whole reason CAMM exists. If they changed it, it would not be CAMM.

The whole purpose CAMM is a thing is to get away for the very limiting design that current RAM modules have where we end up with these really long traces.
I had in mind: " nobody have said that form factor of a desktop computer will not change" :)

I see plenty of options that CAMM will be extremely beneficial for future desktop form factors, which CAMM will enable in itself.
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
I had in mind: " nobody have said that form factor of a desktop computer will not change" :)

I see plenty of options that CAMM will be extremely beneficial for future desktop form factors, which CAMM will enable in itself.

oOOOOooo, gotcha. I misread it as the memory could change for desktop computers. My mistake :)
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,691
136
If you are keeping CPU temperature at 40-50 degrees under load, you are working your cooling system too hard, and your system is actually using more power and producing more heat (and noise) overall. Slow down the fans and let the CPU work at the high end of the temperature range it is designed for. Only increase fan speed if you run into the CPU's temperature limits, affecting performance (boost behaviour, throttling) or stability (when overclocking, undervolting).

Even with the fan almost off, in ECO mode, at full load my 5700X barely breaks 50C. You just have to size your cooling system* right, you can get very low temperatures without overworking anything. This shouldn't be a problem at all in a full size ATX tower.

Now, I agree for more compact mATX and ITX systems it can be a challenge to remove heat fast enough while keeping noise levels down. So there has to be compromises somewhere. But so long as everything stay within specs, there shouldn't be any issues.

*nothing more advanced then an Arctic Freezer 34 tower upgraded with a high pressure Noctua fan for silent operation. Add a couple of slow turning 140mm Noctuas for case ventilation, and heat gets out of the case rather fast.
 

Motostu

Senior member
Oct 5, 2020
497
528
106
Since this thread hasn't been updated in a while, I just wanted to note that the latest ASUS BIOS for my X570 board now allows for the undervolting to be done there instead of having to run PBO2 tuner at each Windows startup. I used the tuner to find the values I wanted to use (I can run -30 on 6 cores, and -25 on the remaining 2), then rebooted and put those values in the BIOS. I used the tool to confirm the offsets were actually working.
 

Makaveli

Diamond Member
Feb 8, 2002
4,717
1,051
136
Since this thread hasn't been updated in a while, I just wanted to note that the latest ASUS BIOS for my X570 board now allows for the undervolting to be done there instead of having to run PBO2 tuner at each Windows startup. I used the tuner to find the values I wanted to use (I can run -30 on 6 cores, and -25 on the remaining 2), then rebooted and put those values in the BIOS. I used the tool to confirm the offsets were actually working.
What did you use to test per core.

Corecycler?
 

Justinus

Diamond Member
Oct 10, 2005
3,173
1,515
136
Yes, and I had one other program I tried, but I think it was primarily core cycler that I used. Once I figured out which cores were good at -30, I modified the config file to only test the 2 remaining cores until I found the offset that worked for them.

Yes, I did it backwards :p
Did you test at default (SSE), AVX2, and AVX512?
 

Motostu

Senior member
Oct 5, 2020
497
528
106
Did you test at default (SSE), AVX2, and AVX512?
Yes. I had prime95 (going through those sets) running on one core while running ycruncher simultaneously on the other (alternating cores).

Once I got it to where it seemed stable I ran it overnight. TBH, since the 2 weaker cores have been stable at -25, I haven't tried to see if they can be tweaked any more. I'm happy with where temps are now.