Jump to content


mark84

Member Since 10 Sep 2008
Offline Last Active Yesterday, 07:15 PM
*****

Posts I've Made

In Topic: Atomic Superpi Scores V5

20 March 2017 - 06:00 PM

Or better yet disable SMT. Something many tests have shown to work better with (because they don't know how to handle SMT properly, like hyper threading, which is the same but I understand implemented differently).

 

I can't do any more benches, had to hand the system back. A reviewers lament.

 

Was fun while I had it.

 

Look forward to seeing what you can cook up Sora! Especially with the higher rated RAM.

These CPUs really do need water cooling though. The air HSF I had couldn't really cope with the higher OCs with all cores enabled. Had to disable 6 as seen in my previous posts so I could push it to 4.25GHz to keep temps in check.

AMDs TDP does not equal Intel TDP ratings. Ryzen pulls as much as Broadwell-E at full tilt. Needs the cooling to match really.


In Topic: Atomic Superpi Scores V5

19 March 2017 - 10:03 PM

Gave wPrime a try again.

 

Ryzen 7 1700X @ 4075MHz w/2400MHz RAM

wPrime 32M 16T 3.7s - 5.2s it wouldn't give me a consistent result. most were in the mid 4s

 

Ryzen 7 1700X @ 3950MHz

wPrime 1024M 16T 96.9s


In Topic: Atomic Superpi Scores V5

19 March 2017 - 10:27 AM

And for comparison, my i7 6850K and near identical clocks of 3.9GHz. Even though it only has six cores, it does have quad channel memory running at 3200MHz.

Manages to beat the Ryzen 7 1700X @ 3.875GHz

 

i7%206850K%203.9GHz%20y-cruncher%201B.jp

 

And when I push the 6850K to the limits (4.5GHz) I get this

i7%206850K%204.5GHz%20y-cruncher%201B.jp

 

A whole 10 seconds quicker...

 

Here's a fascinating insight from the y-cruncher developer about his Zen tuned latest y-cruncher version

 

I went through a lot of trouble to do this in time for Pi day, but here it is. y-cruncher v0.7.2 has a new binary specifically optimized for AMD's Ryzen 7 processors.

 

The performance gain is about 5% over the Broadwell-tuned binary and 15% over v0.7.1. It turns out that the optimizations between v0.7.1 and v0.7.2 happened to be more favorable to AMD Zen than to Intel processors. Nevertheless, this is not enough to make Ryzen beat Haswell-E or Broadwell-E.

 

It's unlikely that any amount of Zen-specific optimizations can make Ryzen beat Haswell/Broadwell-E. The difference in memory bandwidth and 256-bit AVX throughput is simply far too large to overcome. AMD made a conscious decision to sacrifice HPC to focus on mainstream.

 

As for the Ryzen platform itself: It's a bit immature at this point. I went out on launch day to grab the Zen parts. In the end, it took me 3 sets of memory and 2 weeks before I finally found a stable configuration that I could use. From what I've seen on Reddit and various forums, I've been unlucky, but I'm definitely not alone.

 

Slightly more concerning is a system freeze with FMA instructions which appears to be have been confirmed by AMD as a processor errata. Fortunately, the source also says this is fixable via a microcode update. So it won't lead to something catastrophic like a recall or a fix that disables processor features.

 

As for the Zen architecture itself. Here are my (early) observations:

  • The FPU block diagrams that were released about Zen appear to be accurate. The FLOPs benchmark is able achieve 4 128-bit CPU instructions/cycle if there is an equal distribution of FP-adds and FP-multiplies. As expected, FMAs have twice the cost since they need both an FP-add and an FP-multiply.
  •  
  • 256-bit AVX instructions are handled efficiently enough that it seems to be beneficial to use 256-bit instructions when there's no overhead to doing so.
  •  
  • Memory is huge bottleneck. Latencies are very high and dual-channel memory is simply not enough to feed this much computational throughput.
  •  
  • The Ryzen 7 can run FMA4 instructions even though the corresponding CPUID flag is cleared. This is probably to enable some compatibility with code written for AMD Bulldozer while discouraging further use of FMA4. On the other hand, XOP instructions did not get this treatment so they will crash. Ryzen's ability to run FMA4 instructions makes it possible to run y-cruncher's XOP binary without crashing for small computations.

For software developers, compiling code on the 1800X is about as fast as the 5960X at stock clocks. But the 5960X has much more overclocking headroom, so it ends up winning by around 15%. For a $500 processor, the R7 1800X is very impressive.


I've also updated the OP and started to include y-cruncher 1B bench times. Get benching people! :)


In Topic: Atomic Superpi Scores V5

19 March 2017 - 09:02 AM

Cause Rybags seems to into wPrime I tried to run it on my Ryzen review PC. It just fails to launch with an error. Maybe they need to update for Ryzen, dunno. Or I downloaded a dud, but from their official website, I doubt.

Anywho. Time to post this Ryzen's scores up.

The RAM is probably holding it back some, only 2400MHz. But I threw 1.55v at the core (maximum software allows) in an attempt to crack the 10s mark, but I just couldn't do it on this CPU. Needed an extra 25MHz that I just couldn't reach. Oh well.

Decided to do a y-cruncher run too. I'm liking this software more and more. I reckon we ditch MaxxPi in favor of y-cruncher.
In y-cruncher did a 1B decimal point multi-threaded run (So options 0 > 1 > 6, when you run y-cruncher).

Ryzen%201700X%20%204.25GHz%20Super%20Pi.

 

Ryzen%201700X%20%203.875GHz%20y-cruncher

 

Updated CPU-Z for y-cruncher when I realized Sora had posted one with the Ryzen logo in it lol


In Topic: What car do you drive?

19 March 2017 - 08:53 AM

Mate, that car is bloody nice! I have personally been looking at the Ford Focus RS 2016 model, but they reckon the Merc A45 is quicker. I love the colour!

Thanks! :)
Yeah A45 is a tad quicker but for the price the Focus RS is fantastic. You won't be disappointed with an RS's performance. Many also say it's more fun too.