Memory Public presentation Tests

Okay so why was the 2990WX so unsatisfying in a good deal of these tests? Asymptomatic as I alluded to sooner it is 100% down to memory bandwidth, a good deal more so than core to core latency or memory latency.

Hera we can see the uninterrupted memory bandwidth for each processor. You'll observation that the 2990WX is a little land on the 2950X and that's receivable to the added latency the dies without memory controllers incur. It's a 7% drop by bandwidth but that alone doesn't explain the carrying out issues we've seen.

Barely to confirm those results I did likewise examine with AIDA64 and here is the memory copy carrying out, again the 2990WX was down 7% on the but that doesn't explain the miserable performance in the encryption, compression and encryption benchmarks.

For that we need to look at memory bandwidth per core, not for the entire processor just rather the individual cores. Arranging these results aside a single ribbon Beaver State core group we encounter that with just one core active the Ryzen CPUs enjoy tremendous bandwidth.

Forthwith delight note the performance of each core within the CPU is measured individually and result you see here is the average bandwidth across entirely the individual cores. Thus the 2700X and 2950X, both 2nd gen parts deliver the same 29 Gi/s. Then the 1st gen Ryzen parts deliver betwixt 24 - 25 GB/s and then we have the 2990WX at 20 GB/s. This is why we saw a slight drop in total memory bandwidth in the previous test, the margin is amplified Hera exhibit the 2990WX to be well-nig 30% slower as were non limited by the DDR4 retentivity in this instance.

The reason the single center bandwidth is down is due to the fact that 16 of the 32 cores aren't connected directly to the memory and therefore suffer multiplied latent period.

Lastly we see that almost every the Skylake-X parts are limited to just 14 United Kingdom of Great Britain and Northern Ireland/s, though this is less of an supply as 14 GB/s per core is essentially overkill, and here is why.

If we rearrange this graph aside the 'all-togs active' result, the arrangement changes quite second. Now for these results every last CPU cores are actively accessing arrangement memory and we're showing the middling throughput of an individual nucleus. Basically with the CPU running at full steamer in a retention intensive workload, this is the typical amount of bandwidth each essence has at its disposal.

This present is the trouble. The 2950X enjoys a bandwidth of 4.4 GB/s per heart when maxed out and this is why the 14 GB/s we proverb with just a single nitty-gritty active connected the Intel CPU's isn't an issue, since the maximum sustained bandwidth of the Skylake-X CPUs is around 64 GB/s, with 5 cores busy in an extremely memory intensive workload you'Re going to use upward all that bandwidth and once you showtime adding more cores you start to see a drop by efficiency equally they aren't course data fast enough.

Naturally the more cores you have the worst of you're exit to be in this test without increasing the overall memory bandwidth. With octa-channel memory the 2990WX would indeed be able to match the 4.4 GB/s per substance of the 2950X. Simply with just quad-channel memory that figure is halved, well a miniscule ended halved imputable the increased response time indeed it's a bit of a double curse. In the end just shy of 2 GB/s of bandwidth per nucleus just isn't sufficiency and we see the problem this causes when running memory sensitive applications like VeraCrypt for instance. Okay and so ahead we move on to overclocking, power consumption and a couple of new tests, let's quickly go finished gaming performance.