By what mechanism would the multicore performance rise 2x without increased single-core performance? I'm kind of confused. Eliminating intra-CPU concurrency bottlenecks, I guess?
AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
71–80 of 165 posts
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#72By what mechanism would the multicore performance rise 2x without increased single-core performance? I'm kind of confused. Eliminating intra-CPU concurrency bottlenecks, I guess?
Which benchmark or comparison are you referring to?
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#73I have the original Threadripper 1950X overclocked to 4.1ghz and I’ve been pretty happy with it. As a developer I can’t find much reason to upgrade 2 years after its release. Single core improvements (from AMD or Intel) aren’t game-changing. More cores won’t do anything for me (.NET Core, angular, etc) I’ve considered switching back to intel for their 5ghz processors but based on benchmarks, I wouldn’t see anything b…
The biggest upgrade over Threadripper 1950X is that the new ones are no longer NUMA. So performance across the cores will be much more consistent even if the peaks are not that much bigger. If your development involves compiling code then the 2x larger L3 is also going to result in huge improvements to code compilation speed as seen in the Epyc Rome reviews. Example: https://www.phoronix.com/scan.php?page=article&ite…
Wait in what sense are the new Threadrippers no longer NUMA?
Edit: Found the Zen 2 Epyc marketing materials that describe this. Yes, apparently memory access on a single socket is uniform, in that all memory access is indirected over the IO chiplet[1]! This may hurt best-case access latency for NUMA-aware workloads? Just speculating.
It's not like non-local caches go away with uniform DRAM access latencies — unless L1/L2/L3 are also non-local to the core and indirected behind the IO chiplet. Which would be really surprising.
[1]: https://www.servethehome.com/amd-epyc-7002-series-rome-deliv...
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#74Worth noting AMD just announced that the first third-gen Threadripper will come out in Nov with 24 cores, and the planned 16-core mainstream chip (the 3950X) is delayed to November: https://www.tomshardware.com/news/amd-ryzen-9-3950x-delay-la... A 32-core chip is still almost certain to show up, since benchmarks have leaked, and folks have also leaked some pics of 3950X's and packaging, but I guess supply/demand have…
> As the Tom's post notes, server and client chips use the same chiplets, so it could be that most of the higher-binned ones are going to server parts; some are higher-margin (7742 is almost $100/core, vs. 3950X under $50/core)
The 7742 is the "halo" chip, though. Or was, rather, until the Epyc 7H12 was announced.
Other Epycs have much lower $/core. For example the 24-core EPYC 7352 is $1350, making it $56/core. The 16-core EPYC 7282 is actually even cheaper than the 3950X at $650, or $40/core.
No doubt bulk orders are going to get the priority, but the margins may not actually be that different depending on what companies are actually bulk-ordering. The $/core drops pretty quickly even just going down slightly in the stack. The 32c 7452 is $65/core.
And we don't entirely know which aspect of the binning is the limiting factor. If it's just functional cores that's going to be different from if they run at the right frequency/voltage. Epyc is all 225 W or lower. Threadripper 3000 AMD could pretty easily slap a 300W TDP on it and ram voltage through the chips that couldn't cut it at Epyc specs. The 2990WX is, after all, a 250W TDP part. The existing socket is already spec'd for more power capability than the top-end Epyc.
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#75By what mechanism would the multicore performance rise 2x without increased single-core performance? I'm kind of confused. Eliminating intra-CPU concurrency bottlenecks, I guess?
It's not likely but you could do a 32 core 128 thread version.
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#76By what mechanism would the multicore performance rise 2x without increased single-core performance? I'm kind of confused. Eliminating intra-CPU concurrency bottlenecks, I guess?
Which benchmark or comparison are you referring to?
Pretty much the only numbers in TFA:
> The single-core score of 1,275 is pretty much the same as for the current flagship Threadripper 2990WX, ...
> But when it comes to multi-core, the Threadripper 3000's score of 23,015 absolutely destroys the Threadripper 2990WX's score of 13,400, ...
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#77By what mechanism would the multicore performance rise 2x without increased single-core performance? I'm kind of confused. Eliminating intra-CPU concurrency bottlenecks, I guess?
It's not likely but you could do a 32 core 128 thread version.
My understanding is that the underlying CCXs and cores are identical across the Zen 2 product line, and all have exactly 2 threads per core. No?
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#78Earlier quoted context omitted.
The biggest upgrade over Threadripper 1950X is that the new ones are no longer NUMA. So performance across the cores will be much more consistent even if the peaks are not that much bigger. If your development involves compiling code then the 2x larger L3 is also going to result in huge improvements to code compilation speed as seen in the Epyc Rome reviews. Example: https://www.phoronix.com/scan.php?page=article&ite…
Wait in what sense are the new Threadrippers no longer NUMA? Edit: Found the Zen 2 Epyc marketing materials that describe this. Yes, apparently memory access on a single socket is uniform, in that all memory access is indirected over the IO chiplet[1]! This may hurt best-case access latency for NUMA-aware workloads? Just speculating. It's not like non-local caches go away with uniform DRAM access latencies — unless L…
In the definition sense? It's a single memory controller on the IO die & one memory domain. The chiplets don't have their own memory controllers.
By contrast Threadipper 1 & 2 had multiple memory controllers. It was 2x dual-channel controllers. As such they were full on real NUMA, just like a multi-processor system.
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#79Earlier quoted context omitted.
The progress is ok, just incremental. AMD had been playing catch up, until Ryzen/Threadripper. Just for comparison, here are the numbers from my GeekBench 4 run: Single-Threaded: 4,746 Multi-Threaded: 34,586 According to a leaked benchmark, the Threadripper 3000's numbers are: Single-Threaded: 5,519 Multi-Threaded: 68,279 The multithreaded benchmark is 2x, that's a no-brainer since it's likely to have 32 cores vs the…
Honestly, that seems rather low. It looks like a lot of Ryzen 3900X's are doing well over 6000 on Geekbench? Ie. https://browser.geekbench.com/v4/cpu/14151649 Is the Threadripper 3000 test also from Geekbench V4? I noticed they added V5. I don't see why it should be so far behind the Ryzen in single thread... unless the boost isn't working properly or something, could be disabled if it's a test chip.
Yep, it's from GBv4. It could be that the 3900X is clocked higher than the TR3000. Given that they're packing more cores on the chip, it's possible that they can't dissipate heat as well and limit the single-core top-end speed on the TR3000s more than on the Ryzen 3000 series.
Re: AMD Ryzen Threadripper 3000 32-core CPU is more bad news for Intel
#80Earlier quoted context omitted.
I would hope that your production hardware matches your developer / staging / testing hardware. Lets say production is 50% slower than what's tested in staging / developer test cases. Is it the data in production that causes this performance loss? Or is it hardware differences? If you are using Intel tools to debug performance problems on developer / testing stages, you probably want to keep using those Intel tools i…
Said interesting networking optimization is a gaping security home that has already been exploited in the wild.
From my understanding, that vulnerability exists only if RDMA is also enabled.
RDMA, the ability to share RAM as if it were local RAM (through a memory-mapped IO mechanism) across Ethernet is not a common setup. The fact that you can perform cache-timing attacks over RDMA + Intel L3 cache is a testament to how efficient the system is if anything.
Consider this interpretation: RDMA + DDIO is so fast, you can perform cache-timing attacks over Gigabit Ethernet(!!). NetCAT (the "vulnerability" you describe) is proof of it.
Cache-timing / side channel attacks aren't exactly the kind of vulnerabilities that most people think of though. Its kinda cool, but its nothing as crazy as Meltdown / Spectre were.