The AMD Ryzen Threadripper 1950X and 1920X Review: CPUs on Steroids
by Ian Cutress on August 10, 2017 9:00 AM ESTCPU Rendering Tests
Rendering tests are a long-time favorite of reviewers and benchmarkers, as the code used by rendering packages is usually highly optimized to squeeze every little bit of performance out. Sometimes rendering programs end up being heavily memory dependent as well - when you have that many threads flying about with a ton of data, having low latency memory can be key to everything. Here we take a few of the usual rendering packages under Windows 10, as well as a few new interesting benchmarks.
All of our benchmark results can also be found in our benchmark engine, Bench.
Corona 1.3: link
Corona is a standalone package designed to assist software like 3ds Max and Maya with photorealism via ray tracing. It's simple - shoot rays, get pixels. OK, it's more complicated than that, but the benchmark renders a fixed scene six times and offers results in terms of time and rays per second. The official benchmark tables list user submitted results in terms of time, however I feel rays per second is a better metric (in general, scores where higher is better seem to be easier to explain anyway). Corona likes to pile on the threads, so the results end up being very staggered based on thread count.
Corona loves threads.
Blender 2.78: link
For a render that has been around for what seems like ages, Blender is still a highly popular tool. We managed to wrap up a standard workload into the February 5 nightly build of Blender and measure the time it takes to render the first frame of the scene. Being one of the bigger open source tools out there, it means both AMD and Intel work actively to help improve the codebase, for better or for worse on their own/each other's microarchitecture.
Blender loves threads and memory bandwidth.
LuxMark v3.1: Link
As a synthetic, LuxMark might come across as somewhat arbitrary as a renderer, given that it's mainly used to test GPUs, but it does offer both an OpenCL and a standard C++ mode. In this instance, aside from seeing the comparison in each coding mode for cores and IPC, we also get to see the difference in performance moving from a C++ based code-stack to an OpenCL one with a CPU as the main host.
Like Blender, LuxMark is all about the thread count. Ray tracing is very nearly a textbook case for easy multi-threaded scaling. Though it's interesting just how close the 10-core Core i9-7900X gets in the CPU (C++) test despite a significant core count disadvantage, likely due to a combination of higher IPC and clockspeeds.
POV-Ray 3.7.1b4: link
Another regular benchmark in most suites, POV-Ray is another ray-tracer but has been around for many years. It just so happens that during the run up to AMD's Ryzen launch, the code base started to get active again with developers making changes to the code and pushing out updates. Our version and benchmarking started just before that was happening, but given time we will see where the POV-Ray code ends up and adjust in due course.
Similar to LuxMark, POV-Ray also wins on account of threads.
Cinebench R15: link
The latest version of CineBench has also become one of those 'used everywhere' benchmarks, particularly as an indicator of single thread performance. High IPC and high frequency gives performance in ST, whereas having good scaling and many cores is where the MT test wins out.
Intel recently announced that its new 18-core chip scores 3200 on Cinebench R15. That would be an extra 6.7% performance over the Threadripper 1950X for 2x the cost.
Facebook
Twitter
RSS
347 Comments
View All Comments
drajitshnew - Thursday, August 10, 2017 - link
You have written that "This socket is identical (but not interchangeable) to the SP3 socket used for EPYC,".Please, clarify.
I was under the impression that if you drop an epyc in a threadripper board, it would disable 4 memory channels & 64 PCIe lanes as those will simply not be wired up.
Deshi! - Friday, August 11, 2017 - link
No AMD have stated that won;t work. Its probably not hardware incompatible, but they probably put microcode on the CPUS so that if it doesn;t detect its a Ryzen CPU it doesn't work. There might also be differences in how the cores are wired up on the fabric since its 2 cores instead of 4. Remember, Threadripper has only 2 Physical Dies that are active. on Epyc all processors are 4 dies with cores on each die disabled right down to the 8 core part. (2 enabled on each physical die)Deshi! - Friday, August 11, 2017 - link
Wish there was an edit function..... but to add to that, If you pop in an Epyc processor, it might go looking for those extra lanes and memory busses that don;t exist on Threadripper boards, hence cause it not to function.pinellaspete - Thursday, August 10, 2017 - link
This is the second article where you've tried to start an acronym called SHED (Super High End Desktop) in referring to AMD Threadripper systems. You also say that Intel systems are HEDT (High End Desktop) when in all reality both AMD and Intel are HEDT. It is just that Intel has been keeping the core count low on consumer systems for so long you think that anything over a 10 core system is unusual.AMD is actually producing a HEDT CPU for $1000 and not inflating the price of a HEDT CPU and bleeding their customers like Intel was doing with the i7-6950X CPU for $1750. HEDT CPUs should cost about $1000 and performance should increase with every generation for the same price, not relentlessly jacking the price as Intel has done.
HEDT should be increasing in performance every generation and you prove yourself to be Intel biased when something finally comes along that beats Intel's butt. Just because it beats Intel you want to put it into a different category so it doesn't look like Intel fares as bad. If we start a new category of computers called SHED what comes next in a few years? SDHED? Super Duper High End Desktop?
Deshi! - Friday, August 11, 2017 - link
theres a good reason for that. Intel is not just inflating the cost because they want to. It literally cost them much more to produce their chips because of the monolithic die aproach vs AMDs Modular aproach. AMDs yeilds are much better than INtels in the higher core counts. Intel will not be able to match AMDs prices and still make significant profit unless they also adopt the same approach.fanofanand - Tuesday, August 15, 2017 - link
"HEDT CPUs should cost about $1000 "That's not how free markets work. Companies will price any given product at their maximum profit. If they can sell 10 @ $2000 or 100 at $1000 and it costs them $500 to produce, they would make $15,000 selling 10 and $50,000 selling 100 of them. Intel isn't filled with idiots, they priced their chips at whatever they thought would bring the maximum profits. The best way for the consumer to protest prices that we believe are higher than the "right" price is to not buy them. The companies will be forced to reduce their prices to find the market equilibrium. Stop complaining about Intel's gouging, vote with your wallet and buy AMD. Or don't, it's up to you.
Stiggy930 - Thursday, August 10, 2017 - link
Honestly, the review is somewhat disappointing. For a pro-sumer product, there is no MySQL/PostgreSQL benchmark. No compilation test under Linux environment. Really?name99 - Friday, August 11, 2017 - link
"In an ideal world, all software would be NUMA-aware, eliminating any concerns over the matter."Why? This is an idiotic statement, like saying that in an ideal world all software would be aware of cache topology. In an actual ideal world, the OS would handle page or task migration between NUMA nodes transparently enough that almost no app would even notice NUMA, and even in an non-ideal world, how much does it actually matter?
Given the way the tech world tends to work ("OMG, by using DRAM that's overclocked by 300MHz you can increase your Cinebench score by .5% !!! This is the most important fact in the history of the universe!!!") my suspicion, until proven otherwise, is that the amount of software for which this actually matters is pretty much negligible and it's not worth worrying about.
cheshirster - Friday, August 11, 2017 - link
Anandtechs power and compiling tests are completely out of other rewiewers results.Still hiding poor Skylake-X gaming results.
Most of the tests are completely out of that 16-core CPU target workloads.
2400 memory used for tests.
Absolutely zero perf/watt and price/perf analisys.
Intel bias is over the roof here.
Looks like I'm done with Anandtech.
Hurr Durr - Friday, August 11, 2017 - link
Here`s your pity comment.