Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 3
|
![]() |
Author |
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
AMD just released the Ryzen 9 7950X3D (16C/32C) with 144 MB of on-package 3D-stacked cache (128 MB of L3). I was leaning towards building my next PC with the 7950X until I found out they were making their flagship desktop CPU with 3D cache.
----------------------------------------I understand that some WCG applications (MCM?) benefit from more L3 cache available since I've read people talking about cache misses (and having to then use system DRAM). Anyone have a last-generation 5800X3D here or planning on getting one of the 7000 series X3D CPUs? There's definitely a noticeable improvement in most games, but my main focus is on science or productivity: music creation in DAWs, video editing, programming/compiling, video transcoding and encoding into AV1 or H.266, etc. I was planning on running the 7950X in the 105W Eco mode to save heat, electricity, but I like how the 7950X3D natively runs at 120W TDP (so perhaps 150-160W power draw or so).
[Edit 1 times, last edit by hchc at Mar 1, 2023 1:14:13 AM] |
||
|
gibbcorp
Advanced Cruncher Joined: Nov 29, 2005 Post Count: 77 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Did you get this cpu? Even though they are made for gaming they seem really good for power consumption. Number of cores is usually most important for crunching. From what I was told from an Intel engineer at my job, cache stores frequently used instruction sets. So when it gets one it has already done it doesn't have to calculate it again as it is stored in cache which obviously makes things faster. How this would work with crunching would be interesting to find out.
|
||
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I haven't yet built my next desktop PC. Been focusing on getting my rebuilt pfSense setup up and running first and next project is TrueNAS Core and getting my backup and DR ducks in a row. (Have a lot of anxiety over not having an organized storage solution at the moment.) So no beefy new PC yet. Still using the Intel i5-4590 (Haswell) with 20 GB DDR3 as my main and about to switch to a i7 8th Gen laptop with 32 GB DDR4.
----------------------------------------I agree. I'm interested in how the X3D cache affects WCG crunching as well. Could probably test this by loading all 32 threads on a 7950X3D with similar MCM or OPN1 work units and seeing if the X3D chiplet ones finish faster or slower. From benchmarks posted, for productivity tasks like video encoding, the regular 7950X seems to be slightly better than the 7950X3D. By the time I make a decision, Zen 4+ refresh (possibly fall 2023?) or Zen 5 (summer/fall 2024?) might be out lol.
|
||
|
|
![]() |