Perpetual benchmark thread...

[kladner]

Originally Posted by James Heinrich

According to Intel, Processor Base Frequency is 4.0GHz, Max Turbo Frequency is 4.2GHz.
I would guess that Prime95 reads the processor frequency on startup before starting the actual benchmark, and turbo doesn't kick in until the CPU is under load. You could use your favourite monitoring utility (e.g. CPU-Z) to monitor CPU frequency in realtime and see how it changes as you start/run the benchmark.

The first is of course true. I have mine set to sync all cores at multiplier 42x. It still will clock down with no extra-system loading, but the clock rate is very twitchy, and goes full on with minimal load increase.

I will have to see when it goes to max when starting the benchmark. This will include trying to limit other loads (like shutting down multi-tabbed Firefox,) and maybe even mfaktc to limit whatever memory contention might arise. Of course, too, limiting other loads is not a Real Life^® situation, either. Most of the time I expect this machine to do a bunch of other stuff when I want it to, and this certainly takes a toll on the ms/it.

[db597]

So from the benchmarks it looks like 8 Ryzen cores is still slower than 4 Skylake/Kabylake cores:

Ryzen @ 3.3GHz:

Timings for 4096K FFT length (8 cpus, 1 worker): 6.92 ms. Throughput: 144.58 iter/sec.

i7-6700K @ 4.2GHz:

Timings for 4096K FFT length (4 cpus, 1 worker):  4.07 ms.  Throughput: 245.91 iter/sec.

Given it's a new architecture, I wonder if there is AMD specific optimisation that can be done. Going above 4 cores on Ryzen yields very small gains. It could be the slow communication between the CCX. If we can keep each worker within it's own CCX, perhaps running 4 cores x 2 workers might be the most optimum for Ryzen.

[Mark Rose]

The @ 4.0GHz is part of the model name when queried from the processor. It has nothing to do with the actual running processor frequency. It has confused me in the past as well.


Kieren, are you running 12 GB of RAM? Kind of an odd amount. Not having matched sticks is probably hampering performance. That being said, you're getting 32% more throughput with a 27% higher CPU clock and a 50% higher memory clock compared to my systems (for 4 cores, 1 worker, 4096K FFT). That extra memory bandwidth is helping.

Your benchmark also tells me I'm still memory constrained at 2133 with 4 cores at 3.3 GHz. I may try poking around the bios to see if there's a way to under a locked CPU besides disabling turbo.

[kladner]

Originally Posted by Mark Rose

The @ 4.0GHz is part of the model name when queried from the processor. It has nothing to do with the actual running processor frequency. It has confused me in the past as well.


Kieren, are you running 12 GB of RAM? Kind of an odd amount. Not having matched sticks is probably hampering performance. That being said, you're getting 32% more throughput with a 27% higher CPU clock and a 50% higher memory clock compared to my systems (for 4 cores, 1 worker, 4096K FFT). That extra memory bandwidth is helping.

Your benchmark also tells me I'm still memory constrained at 2133 with 4 cores at 3.3 GHz. I may try poking around the bios to see if there's a way to under a locked CPU besides disabling turbo.

Does it indicate 12 GB somewhere? It should be 16GB, dual channel, dual rank. They are rated at 2666MHz, running at 3200.

[VictordeHolland]

Originally Posted by db597

So from the benchmarks it looks like 8 Ryzen cores is still slower than 4 Skylake/Kabylake cores:

Ryzen @ 3.3GHz:

Timings for 4096K FFT length (8 cpus, 1 worker): 6.92 ms. Throughput: 144.58 iter/sec.

i7-6700K @ 4.2GHz:

Timings for 4096K FFT length (4 cpus, 1 worker):  4.07 ms.  Throughput: 245.91 iter/sec.

Ryzen is about as fast as my 5?6? year old SandyBridge

i5-2500k @4.0GHz DDR3-2133

Best time for 4096K FFT length: 6.839 ms., avg: 7.155 ms.
Timings for 4096K FFT length (4 cpus, 4 workers): 27.12, 26.85, 27.58, 27.00 ms.  Throughput: 147.41 iter/sec.

[Mark Rose]

Originally Posted by kladner

Does it indicate 12 GB somewhere? It should be 16GB, dual channel, dual rank. They are rated at 2666MHz, running at 3200.

Yeah, it does:

 Machine#0 (total=12649168KB, Backend=Windows, hwlocVersion=1.11.6, ProcessName=prime95.exe)

I doubt it will affect Prime95 though, now that you've confirmed the install RAM.

[James Heinrich]

Originally Posted by Mark Rose

The @ 4.0GHz is part of the model name when queried from the processor. It has nothing to do with the actual running processor frequency.

I don't think it's the "@ 4.00 GHz" that was concerning, it was the "CPU speed: 4008.14 MHz".

Originally Posted by Mark Rose

are you running 12 GB of RAM? Kind of an odd amount. Not having matched sticks is probably hampering performance.

That depends on the system configuration. My i7-920 system has 12GB, but it's triple-channel so 12GB is a balanced configuration. Granted most systems are dual-channel (I'm odd and my two systems are 3-channel [i7-920] and 4-channel [i7-3930K] )

As for the RAM reported, I believe that's what's available to Prime95, not total system RAM. In my case I have 64GB installed and it logs as

Machine#0 (total=54609356KB)

which is 52GB. Although I'm not entirely sure how it pulled up that number, since I have 5 workers, specified at 11000MB each, maximum 4 high-memory workers, and an overall maximum of 44000MB. But in any case, it's clearly not the installed system RAM amount.

[kladner]

I don't think it's the "@ 4.00 GHz" that was concerning, it was the "CPU speed: 4008.14 MHz".

I see numbers like that when running stock. I think it must come from variations in the base clock.

[James Heinrich]

Originally Posted by kladner

I see numbers like that when running stock. I think it must come from variations in the base clock.

It's not that it wasn't exactly 4000.00MHz, but rather that kladner was expecting ~4.2GHz, not ~4.0GHz, hence my suggestion to monitor the frequency in realtime.

[kladner]

Originally Posted by James Heinrich

It's not that it wasn't exactly 4000.00MHz, but rather that kladner was expecting ~4.2GHz, not ~4.0GHz, hence my suggestion to monitor the frequency in realtime.

I did try watching CPU-Z when starting the benchmark. It takes thinning out other CPU users to get a baseline. As mentioned, in the "Sync all cores" option on the Asus board seems to make the frequency (multiplier) jump around a lot. When things were quiet, and the core clock was only occasionally hitting 4200 MHz, moving the mouse of clicking on something would make it peak.

The jump to 42x seems virtually simultaneous with clicking to start the benchmark, at least to human-scaled perceptions.

[FSund]

Originally Posted by db597

The 8192K FFT performance looks incredible on this version of Prime95, especially when all 8 cores are thrown at it. Would be good if someone can post results from a similarly priced Intel i7 7700K on Prime95 v29.1 Build 15 for comparison (I expect the i7 is a lot faster per core, but at the end of the day having double the cores may make it a rather close competition).

I have just gotten a 7700k, and I'm in the process of overclocking it now. Will try to report back with benchmarks when I'm done.

At the moment it's looking like I'll have to accept 4.7 GHz for Prime95 runs, any higher and my temperatures get too high. Or to be more specific, at 4.7 GHz have to increase the voltage to 1.280 V to do the the Prime95 torture tests without errors, and at those voltages I get peak temperatures of 85 C, which is a bit too high for my comfort.