New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Personally I find it interesting still as it's relative testing and well it's better than most benchmark tests that test on windows rather than linux. It's better than having no comparative metrics to go by as an indicator of relative performance improvement and/or regression between cpu generations.
True, though if they aren't focused on performance they could go down one cpu model for AMD EPYC 7282.
AMD EPYC 7282 2nd gen 2.8Ghz base and 3.2Ghz turbo with 16C/32T 120W
with 8MB L2, 64MB L3 and DDR4-3200Mhz
at $650 RRP but retail pricing is around $770-790 mark which is around same price as AMD EPYC 7351P 1st gen.
Would still be faster than all those old Intel Xeon E-2630-2680 v1/v2/v3/v4 that some LET providers are selling still but the AMD EPYC 7282 would still be 2x price of those Intel Xeons if you're sourcing them off ebay heh
Though with some Intel vulnerabilities like MDS https://www.intel.com/content/www/us/en/architecture-and-technology/mds.html and requirement to disable hyperthreading, you will loose 1/2 your cpu thread count compared to with AMD EPYC or newer Intel 8/9th gen desktop or newer Cascade Lake and some select Intel Coffee Lake cpu models which are NOT affected by MDS vulnerabilities. Trust me the software/kernel based MDS mitigation performance hit/overhead on Intel vulnerable cpus is real compared to benchmarks I did against AMD EPYC and even Intel newer cpus with hardware mitigated MDS fix. And that is without Hyperthreading disabled !
So the maths equal out. Older Intel Xeons sourced on ebay would technically require disabling hyperthreading to fully mitigate MDS so 1/2 the price for 1/2 the performance in some work loads that had benefit for hyperthreading versus AMD EPYC 7282 at twice the price but >2x the number of cpu threads - meaning hosting more LET targeted VPSes on same server.
Example dual Intel Xeon E5-2680v2 at 2.80ghz base is 10C/20T x2 so 20 cores and 40 threads but with HT disabled has 20 cores for full MDS vulnerability mitigation.
versus AMD EPYC 7282 at 2.80ghz base dual cpu would have 16C/32T x2 so 32 cores and 64 threads - 3.2x times more than E5-2680v2
even a single AMD EPYC 7282 would have 16C/32T so 32 threads versus disabled HT E5-2680v2's 20 cpu cores is 1.6x more.
Sort of makes sense seeing alot of Linode's new VPS nodes sporting AMD EPYC cpus - though 1st gen right now. Hoping to see AMD EPYC 2nd gen later
@eva2000
That's all nice and true but usually the point for a provider isn't processor cost but rather the cost of the complete system. And there no matter which Gen2 Epyc they'd need to buy a new system costing multiple times of what they'd have to pay for a 2nd hand Gen1 system.
True indeed. Relatively newer released cpus requiring new systems would cost more. LET and ultra lowend VPS hosting is a tricky mathematics balance I can imagine