|
I'm pretty impressed with the Ryzen performance for business purposes. I think it's the first time in years I've been enthusiastic about a cpu. Seems like an afforable way to get a lot of cpus with better floating point performance than intel.
|
#
¿
Aug 15, 2017 04:27
|
|
|
|
| # ¿ Apr 29, 2024 02:39 |
|
|
AVeryLargeRadish posted:That makes a lot of sense, Ryzen is the first compelling new thing we have seen in desktop CPUs since Sandy Bridge or so. Hopefully this lights a fire under Intel and we see some real innovation from both sides over the next few years. My workload is CFD so it's pretty easy to create a mesh for each virtual cpu and of course what I run is so archaic that it uses Fortran so every node has at least 10 floating point variables. A 1950X could theoretically reduce the run times by 75% compared with my quad core xeon. There's a big cost jump between an 1800X and a 1950X (including the cost of the motherboard) so I'll have to think about this for a while.
|
|
#
¿
Aug 15, 2017 04:48
|
|
|
Eyochigan posted:If you're spending money, make it worthwhile. You could save money now, or save time later. Is it worth upgrading from what you have to an 1800x? Why not a 1700? A good question. All of the CPUs I'm using for work and home (excluded my laptop) are all getting rather old (the memory in my workstation is DDR3 - I'm assuming this is rubbish by today's standards). An 1800x would be a major upgrade and with the fast floating point could be a leap forward that's faster than what I'm expecting. Why not a 1700? I am prepared to pay for higher clockspeed and prefer not to overclock. Overclocking on a multithreaded job that could take up to a week to complete is a commercial risk if it's unstable.
|
|
#
¿
Aug 16, 2017 05:28
|
|
|
Mofabio posted:Could I make an alternative suggestion for your CFD workload? Well it's worth thinking about. People have commented about memory bandwidth. For most CFD number crunching the CPU itself is usually the bottleneck as everything chokes up with the floating point calculations. When I was comparing the floating point throughput between Ryzen and the recent Xeons Ryzen is about 30% faster at the same clockspeed. There is a possibility that I might be wrong as some of the simulations I run might be broken down into such small workloads that the bottleneck may shift to cache or memory in whole or part. Although I have my doubt as the larger the model the complexity is exponential in terms of calculations required. I'll have to think about this a bit more and I'm not closing out any options.
|
|
#
¿
Aug 17, 2017 03:07
|
|
|
I got a 1700X for home office use. Under the typical office use of rendering 3d fractals and playing GTA V it's fantastic.
|
|
#
¿
Sep 1, 2017 03:37
|
|
|
So how are we supposed to benefit from the 128 pcie lanes on Epyc? What sort of performance could be squeezed out of the hardware?
|
|
#
¿
May 10, 2018 01:47
|
|
|
GRINDCORE MEGGIDO posted:Split out the pcie lanes and mine with 100+ cards at once, this is a fantastic idea. I could see 128 threads in Epyc being used to produce bulk dried strawberries.
|
|
#
¿
May 10, 2018 05:45
|
|
|
I watched that last night. From the building work that I do those old radiators are mostly rust with a little bit of metal. I wasn't surprised to see the state of the water after heating that storage room.
|
|
#
¿
May 10, 2018 23:07
|
|
|
The CFD software I use includes a lot of combustion and thermal radiation calculations. At the start of the simulation it is usually heavily memory bound until there is a lot of movement in the model then it tends to be heavily cpu bound. At least I think that's correct as recently I found out the cpu load reporting includes treating a cpu as busy with cache miss and the cpu is halted waiting for a memory read.
|
|
#
¿
Jun 5, 2018 22:59
|
|
|
Dadbod Apocalypse posted:Haha, the idea of possibly owning a 32-core PC is mind blowing. I didn't replace my quad core xeon workstation as I was waiting to see what competition between intel and amd would deliver. I'd be quite happy with a 32 core workstation mostly idle running word and for 2d drawing (with occassional CFD workloads running for days).
|
|
#
¿
Jun 6, 2018 05:03
|
|
|
The price of the Epyc chips is pretty drat crazy at the low end so yeah 32 core TR is likely to have an eye bleed price.
|
|
#
¿
Jun 6, 2018 06:43
|
|
|
Where I'm finding Zen 2 interesting is in the doubling of the floating point instruction pipelines. When I run fluid dynamics simulations it's all cpu bound and primarily floating point instructions. The difference between my old xeon workstation and my 1700 at home is a 60% increase in speed by doubling cores (even though ryzen is a bit slower in IPC and clock speed). I'm prepared to wait for something Zen 2 with at least 8 cores for my new workstation.
|
|
#
¿
Nov 10, 2018 20:17
|
|
|
Pablo Bluth posted:Whats the Xeon configuration? With CFD you pretty quickly run in to being memory bound. So quad channel Threadripper should beat any Ryzen based solution, and Epyc will again be faster due to having eight memory channels. Given eight memory channels is ahead of Intel's offering of six, there's been a lot of interest in Eypc for HPC uses, unfortunately like-for-like benchmarks have been thin on the ground. It's an old xeon but for the models I'm running I'm not hitting any memory bound issues. I'm running FDS and the fire calculations end up very intense. In the past I'm only seen memory become a problem on a very large model at the start of the run but once the fans start in the model the cpus would always sit at 100%. Even NIST who wrote the software suggest only running as many threads as cpu cores because memory latency isn't an issue.
|
|
#
¿
Nov 12, 2018 06:10
|
|
|
Pablo Bluth posted:As I understand it, the usual CPU utilisation number shown doesn't tell if the CPU is memory bound. A CPU is 'busy' whether it is performing instructions or waiting for data from memory (because it's not in the CPU caches). Any CPU utilisation drop at the start of the run is usually a disk IO bottleneck when large runs can't be feed in to memory quickly enough. I'd eliminated disk bottleneck so that wasn't an issue as the model loads entirely into ram. In terms of cpu utilisation I've seen a talk by a coder for Netflix in relation to that. From testing I've done the halt while waiting for cache misses doesn't seem to be an issue either. e: AMD have suggested mixed integer and fp workload ipc is increased by29%. Devian666 fucked around with this message at 19:53 on Nov 12, 2018 |
|
#
¿
Nov 12, 2018 19:46
|
|
|
BangersInMyKnickers posted:Anybody here running Epyc servers after coming off Xeons? I have a core-hungry workload and the benchmarks I'm seeing say I can get roughly 50% more IOPS for my dollar but it would be nice to talk to someone who has made the switch about any caveats that have come up. I can't even find any Epyc servers in my region. Waiting for a 12 core Zen 2 Ryzen to replace my workstation at the moment, which won't be released until around March next year.
|
|
#
¿
Dec 11, 2018 02:40
|
|
|
SwissArmyDruid posted:Ah yes, longer and deeper pipelines per clock cycle, where have I heard this before... I remember those days of the inferno of heat coming out of the Pentium 4 and CRT monitor. So much heat for a 2.4 GHz single cpu with hyperthreading. The temperature in the office was intolerable back then. I'm waiting to see how Zen 2 compares with the new intel cpus. Especially given that intel seems to be secretly adding an extra AVX pipeline to some of their cpus.
|
|
#
¿
Dec 18, 2018 23:08
|
|
|
|
| # ¿ Apr 29, 2024 02:39 |
|
|
PC LOAD LETTER posted:AVX (especially the 512bit version) isn't used all that much though still and probably will never be. Developers just don't seem to be as interesting in pushing it into software everywhere as they were with SSE2 for instance. 256bit AVX does seem to be getting some more use but its still being slow rolled unfortunately quite a bit. Unless you've got something specifically that uses it then it doesn't seem to be all that much to be excited about right now as far as CPU features go. I do have some software that uses AVX/AVX2 and might be able to compile for AVX-512. With Zen 2 doubling the pipeline for AVX2 buying intel doesn't provide much difference now. AMD having Zen 2 with a higher core count with high clock speeds is likely to win me over for my next workstation purchase. Just waiting to see what the official announcement is.
|
|
#
¿
Dec 19, 2018 01:11
|
|