Thanks for the post Alain. Was nice to hear from someone who isn't a gamer also. One of the cpu benchmarks is a chess based test and Ryzen comes out in good shape because of the abundance of cores. 2933 is a useful memory speed and it takes a lot more dollars to upgrade to better memory for faster speeds. But there is a very measureable difference in compute speeds when you get your memory subsystem clocked higher. I've got both my Ryzen's running at 3333 Mhz at CL14 with tight timings. I do distributed processing as my passion and I shaved a couple of minutes off my run times per task with the bump from 3200 to 3333. So some further tuning is possible if you want to pursue that.
Hey now, I do more than gaming. I do video editing and live streaming, so I'm not just a one trick pony... I know your remark wasn't singled at me, but I know the benefits of RAM speed for other things than just gaming performance. Which is a thing the YT and web reviewers focus too much on! They just called Ryzen '****' because it wasn't better than Intel at gaming, nvm the fact it HOSES Intel in other areas. I've literally had my R5 1600 at the top of AIDA test lists in certain benchmarks, out doing their 6 core HEDT processors on X99... Yeah, X99 is a bit old, but Ryzen is ALL-NEW architecture and it's in it's infancy.
With Zen+ being on a 10NM process we can expect at least 10% performance. That means CPU's should hit 4.4 or 4.5 on OC, and hopefully they get some IPC gains, and WHAM, you are getting damn close to Intel performance after one refresh of the Zen architecture.
Sorry that I get carried away, but I'm not an Intel fan. Their shady business practices back in the day almost destroyed AMD, and I'm all about fair competition and not throwing your weight around because you dwarf your competitor in finances.
If AMD can get Zen+ and Zen 2 to equal or match Intel, then it's pretty much AMD winning big-time. AMD can make their processors for a significantly lower cost than Intel because of Infinity Fabric technology. The days of huge monolithic dies are numbered. If AMD puts enough heat on Intel then they'll have to adopt manufacturing the way AMD is. Hell Intel already has their own version of Infinity Fabric (I forget the name at the moment).
Anyhow, I decided on 2800 CL12 1.35v CR1 RAM speed as 2933 was all I could get with CL14 1.385v CR2. I figured 2800 with lower timing and CR1 with lower volts would be about the same performance or better anyways. Synthetic benchmarks say the RAM is just as fast in read, write, and copy as it was with 2933. The system loads in 22 seconds from cold boot, and programs are really snappy.
Going back to gaming, I didn't see any increase in performance. This is probably because my R9 390 is being fed well enough without the increased RAM speeds. So when reviewers only test with a 1080 Ti they are not even remotely covering the gambit of GPU's out there and making such blatant statements is causing everyone to think they NEED higher speed RAM. That is not the case. Your configuration of hardware, software, and what apps you are running will determine your need for RAM speed and what is optimal. Also, don't forget about IMC, that will affect your max memory OC a lot if the IMC is not that great. SO many factors go into the RAM timings and what is going to see benefit from higher speed RAM anyways.
Thanks Keith,
3333mhz that's nice... I bought the Ripjaws V that is not on the QVL list because it was the only one available in my store and i didnt want to wait
. I will try to push manually but i don't have any esperance after all i read about this memory.
but i'am still satisfied with 2933 mhz.
Thank you
Alain
Alain, I have no idea what your program needs, but I suspect since it's a HUGE data base, hard disk speed would seem more important for finding the data than RAW RAM speed performance. It's pure speculation on my part, since I don't know what your program needs and what exactly it does.
I apologize for going off tangent, but I think about these things a lot since I review PC hardware.
