Hacker News new | ask | show | jobs
by Symmetry 2872 days ago
Rendering is something I could see people doing at home if that's their hobby.
1 comments

Then the tests in this article are hardly convincing.
How is the $1800 2990WX outperforming the $1980 i9-7980XE in every rendering test they performed not convincing?

There are clearly workloads where the 32 core TR chip does not perform well (probably due to the memory configuration) but it seems pretty good at rendering.

The 2990wx blew everything out of the water in rendering. It's like 37% faster than the 7980xe in the Blender benchmark.
Faster with 2x more cores and twice more power consumption. Hardly a win in my book - it means each core is way weaker and power hungry. Intel will easily match that sometimes soon without sweating.
Check your numbers. In that Blender benchmark, AMD is actually 58% faster (152/96), has 78% more cores (32/18), and its TDP is 52% higher (250/165).¹ https://www.anandtech.com/show/13124/the-amd-threadripper-29...

This means AMD manages to execute more work per watt (more energy efficient), and each AMD core uses less power than Intel.

¹ Anantech wrongly lists the TDP as 140W. It's in fact 165W: https://ark.intel.com/products/126699/Intel-Core-i9-7980XE-E...

A Corona 1.3 benchmark between the i9-7980XE and the 2990WX saw, in that workload, the 2990WX 28% faster than the i9.

https://www.youtube.com/watch?v=QI9sMfWmCsk&feature=youtu.be...

Its power consumption was 19% higher than the i9-7980XE.

https://www.youtube.com/watch?v=QI9sMfWmCsk&feature=youtu.be...

Tom's Hardware saw a stock 2990WX at a lower power consumption than a stock i9-7980XE during a Prime95 "torture loop". Overclocked, the AMD part was higher than the Intel one, but only slightly.

https://www.tomshardware.com/reviews/amd-ryzen-threadripper-...

Where have you seen that it has double the power consumption? Under what workloads?

Personally, I don't care about "weaker cores". If a system has 2048 cores clocked at 7 THz and it is 20% faster at my workload than a single-core CPU at 700 MHz, it is faster.

The fact that the "weaker cored" system is cheaper than the "burly muscly" single core system is a bonus.

Power consumption doesn't even matter that much either. It is the equivalent to a single 60W light bulb (or several of those new-fangled LED bulbs). Big whoop.

If I were Pixar and was running one of these 24/7 then, well, it's theoretically possible that the extra you pay in electricity would make up for the lower capital price. But for a hobbyist running this at most 10 hours a week I really doubt that that's a consideration.

But more importantly, The Tech Report looked at task energy for the Threadripper in rendering tasks and found that it took less energy to finish a render than competitorys. It's power was higher but the time was shorter to an even greater extent.

https://techreport.com/review/33977/amd-ryzen-threadripper-2...

So if you're so serious about rendering that you're willing to spend thousands on a good rig for it there really isn't any reason not to use this boy.

If you check Phoronix, they found it doesn't actually take much more power than the 7980XE.
2x core count hardly ever offers anywhere near a 100% performance gain, even across Intel's lineup [1]. Very few workloads are that parallelizeable, and almost everything (from the program to the OS to the CPU itself) introduces some form of overhead when running in parallel.

[1] https://www.cpubenchmark.net/compare/Intel-Core-i9-7960X-vs-...

How much power would an Intel based system need to get that 40% performance boost?
What metric are you considering? The rendering tests show the 2990WX to be both faster and cheaper than the i9.
Slightly faster yes. Cheaper? For now. Price is elastic. The i9 could be priced at any price point because there was no competition until now. Do you think Intel will keep it overpriced for long?
AMD has manufacturing cost advantage - CPUs are built from 4- core CCXes that can be binned separately.

Intel on the other hand needs to manufacture a monolithic CPU that not only is fault free in enough cores, but performs well. That's harder and yields are way lower.

80% yield on a 4 core block is a 16.7% yield on a 32 core block - and that's before binning

If AMD's method scales so much easier, why didn't Intel just... do that? Honest question.
Intel could do that. But AMD came out with this method first.

AMD has only been doing this "infinity fabric" thing for a year. Intel was caught with their pants down. It seems like Intel is researching chiplet technology and trying to recreate AMD's success here.

It takes several years to create chips. So Intel realistically won't be able to copy the strategy until 2020 or later. But you better bet that Intel is going to be investing heavily into chiplet technology, now that AMD demonstrated how successful it can be.

Intel has done that in the past, actually, their first "dual core" chip (2005) was actually two chips in a package.

https://en.wikipedia.org/wiki/Pentium_D

Because it's not free. Communication between cores in different CCXes (and memory access - there is a single memory controller per CCX) has slightly higher latency than within a single CCX (or a monolithic CPU, but here Intel's advantage decreases with core count due to a different interconnect).

Also because they didn't have to innovate - no competition since early Opterons.