An EPYC NixOS Build Farm

#1

Post by nuriaion » Sat, Aug 18, 2018, 10:20 AM UTC

Re: An EPYC NixOS Build Farm

#2

Post by tmd83 » Sat, Aug 18, 2018, 2:46 PM UTC

This is quite surprising. I assumed they would get an EPYC server with higher core at similar price (due to the cost of the CPU) and get advantage in performance/dollar. I didn't expect EPYC to beat a similar physical core count intel server running at similar core count let alone by this much margin. Am I missing something here?

I'm saying this as someone who really wants AMD to succeed and push the ecosystem forward. I just want to make sure that I'm reading the results correctly here.

They are effectively getting 35% extra performance 60% of the price that's just beyond impressive.

Re: An EPYC NixOS Build Farm

#3

Post by boris » Sat, Aug 18, 2018, 2:58 PM UTC

tmd83 wrote: post #2

This is quite surprising. I assumed they would get an EPYC server with higher core at similar price (due to the cost of the CPU) and get advantage in performance/dollar. I didn't expect EPYC to beat a similar physical core count intel server running at similar core count let alone by this much margin. Am I missing something here? I'm saying this as someone who really wants AMD to succeed and push the ecosystem forwar…

EPYC is compared to a previous generation Xeon with lower TDP (105w vs 155/170w) and slower RAM (2400 vs 2666). The Xeon server is more expensive probably because it has 4x the amount of RAM.

Edit: actually, this is a dual Xeon machine so its TDP is slightly higher than EPYC's.

Re: An EPYC NixOS Build Farm

#4

Post by phkamp » Sat, Aug 18, 2018, 3:31 PM UTC

tmd83 wrote: post #2

This is quite surprising. I assumed they would get an EPYC server with higher core at similar price (due to the cost of the CPU) and get advantage in performance/dollar. I didn't expect EPYC to beat a similar physical core count intel server running at similar core count let alone by this much margin. Am I missing something here? I'm saying this as someone who really wants AMD to succeed and push the ecosystem forwar…

The EPYC has twice as many memory busses.

Re: An EPYC NixOS Build Farm

#5

Post by yjftsjthsd-h » Sat, Aug 18, 2018, 4:06 PM UTC

boris wrote: post #3

tmd83 wrote: post #2

This is quite surprising. I assumed they would get an EPYC server with higher core at similar price (due to the cost of the CPU) and get advantage in performance/dollar. I didn't expect EPYC to beat a similar physical core count intel server running at similar core count let alone by this much margin. Am I missing something here? I'm saying this as someone who really wants AMD to succeed and push the ecosystem forwar…

EPYC is compared to a previous generation Xeon with lower TDP (105w vs 155/170w) and slower RAM (2400 vs 2666). The Xeon server is more expensive probably because it has 4x the amount of RAM. Edit: actually, this is a dual Xeon machine so its TDP is slightly higher than EPYC's.

Doesn't that make it more impressive? The AMD build has a quarter of the RAM and still came out ahead!?

Re: An EPYC NixOS Build Farm

#6

Post by wereHamster » Sat, Aug 18, 2018, 4:15 PM UTC

yjftsjthsd-h wrote: post #5

boris wrote: post #3

Earlier quoted context omitted.

EPYC is compared to a previous generation Xeon with lower TDP (105w vs 155/170w) and slower RAM (2400 vs 2666). The Xeon server is more expensive probably because it has 4x the amount of RAM. Edit: actually, this is a dual Xeon machine so its TDP is slightly higher than EPYC's.

Doesn't that make it more impressive? The AMD build has a quarter of the RAM and still came out ahead!?

Unsurprisingly, the amount of RAM has little effect on a CPU-heavy workload.

Re: An EPYC NixOS Build Farm

#7

Post by stefan_ » Sat, Aug 18, 2018, 4:25 PM UTC

yjftsjthsd-h wrote: post #5

boris wrote: post #3

Earlier quoted context omitted.

EPYC is compared to a previous generation Xeon with lower TDP (105w vs 155/170w) and slower RAM (2400 vs 2666). The Xeon server is more expensive probably because it has 4x the amount of RAM. Edit: actually, this is a dual Xeon machine so its TDP is slightly higher than EPYC's.

Doesn't that make it more impressive? The AMD build has a quarter of the RAM and still came out ahead!?

For this workload, they could have had 16 GiB of RAM and get pretty much identical results.

Re: An EPYC NixOS Build Farm

#8

Post by bearjaws » Sat, Aug 18, 2018, 4:47 PM UTC

For anyone confused by instance types, this is for Packet.net

So these two servers are pretty similar, I originally thought they were comparing a Amazon m1.xlarge vs EPYC...

Re: An EPYC NixOS Build Farm

#9

Post by chris_st » Sat, Aug 18, 2018, 5:00 PM UTC

I'm curious if anyone knows the test tool they're using here, in particular for lines like:

  $machine->waitForWindow("^Desktop ");

It looks incredibly useful.

Thanks!

Re: An EPYC NixOS Build Farm

#10

Post by O_H_E » Sat, Aug 18, 2018, 5:31 PM UTC

chris_st wrote: post #9

I'm curious if anyone knows the test tool they're using here, in particular for lines like: $machine->waitForWindow("^Desktop "); It looks incredibly useful. Thanks!

Well there seems to be something called nix-shell, but I never heard of it before. It seams pretty useful. The link to plasma tests is provided in the article.

https://github.com/NixOS/nixpkgs/blob/49a6964a4250d98644da61...