Exclusive: Intel Xeon D-1518 Benchmarks

[Patrick Kennedy]

Today we have Intel Xeon D-1518 benchmarks. Our most popular topic in 2015 was Intel Xeon D and in November 2015 we heard about the Intel Xeon D wave 2 launch and we have been pushing to benchmark the new chips

The post Exclusive: Intel Xeon D-1518 Benchmarks appeared first on ServeTheHome - Server and Workstation Reviews.

Related posts:

1. Exclusive: Intel Xeon D-1520 benchmark results
2. Exclusive: Intel DC S3610 400GB Quick Benchmarks and Mini-Review
3. Exclusive: Gigabyte – Annapurna Labs ARM storage server benchmarks
4. Intel Xeon E3-1240 V3 Review – Haswell Xeon Benchmarks

Continue reading...

 

[gigatexal]

Nice writeup. Do you have a 2670 testbed around? It'd be interesting to see how these once 1.5k chips stack up against the newer broadwell-d chips etc.

But still, nice write-up, looking forward to the D-1528 review.

 

[Patrick]

Nice writeup. Do you have a 2670 testbed around? It'd be interesting to see how these once 1.5k chips stack up against the newer broadwell-d chips etc.

But still, nice write-up, looking forward to the D-1528 review.

I can do that without too much effort given what is in the Sunnyvale DC. Those chips are 8C/ 16T each so I did not think 16C/ 32T was a good comparison against a low clock speed 4C/8T.

 

[Deslok]

What about any L5420 comparisons? those at least were everywhere at one point

 

[Patrick]

What about any L5420 comparisons? those at least were everywhere at one point

Great point, just no comparison hardware.

 

[Deslok]

Great point, just no comparison hardware.

I'll have to check I have too many Dell sc24-cs systems at work but I don't know if I can take one offline... One is a hyper v host I might be able to migrate the guests off of it and do a usb boot temporarily though nothing I could ship to you unfortunately

 

[Patrick]

The other side to it is you can use Ubuntu 14.04.3 and just do a comparison yourself. It would probably take 55-70 min for Linux-Bench to run once on that machine.

 

[Diavuno]

Great point, just no comparison hardware.

I have a few 53xx and 54xx boxes.

I know a handful are indeed the L5420

 

[gigatexal]

I can do that without too much effort given what is in the Sunnyvale DC. Those chips are 8C/ 16T each so I did not think 16C/ 32T was a good comparison against a low clock speed 4C/8T.

of course, they are in much different leagues, I totally agree @Patrick , but a lot of us have 2670s now and I was until recently looking at getting a Xeon-D, would be nice to see how it stacks up just for comparison sake.

 

[smithse79]

One thing I noticed but wasn't mentioned: The C2758 is on par or just slightly behind the 1518, but with only a 20W TDP. I find it interesting that the atom still is probably better / watt than the Xeon-D

 

[lucidrenegade]

One thing I noticed but wasn't mentioned: The C2758 is on par or just slightly behind the 1518, but with only a 20W TDP. I find it interesting that the atom still is probably better / watt than the Xeon-D

I noticed that too. The only instances where the 1518 wins by a good margin are the OpenSSL and whetstone benches. Seems that the extra 4 cores in the C2758 really help (at least in the benchmarks Patrick ran).

 

[Deslok]

The 2758 is a monster chip but performance/watt might be in favor of Xeon-D at the high end it moves from 35-45watt when you move up to a 154x chip

 

[Patrick]

I noticed that too. The only instances where the 1518 wins by a good margin are the OpenSSL and whetstone benches. Seems that the extra 4 cores in the C2758 really help (at least in the benchmarks Patrick ran).

I think there are a few other highlights:

• D-1518 is significantly faster in single threaded benchmarks
• The D-1518 is a much better platform. 2x 10Gb v. 4x 1Gb, 6x SATA3 v. 2x SATA3 + 4x SATA2, 128GB RAM v. 32GB (ok 64GB but practically 32GB given available SODIMMs), PCIe 3.0
• The C2758 has the onboard crypto engine. The Xeon D requires Coleto Creek PCIe/ PCH for QAT. So I do think there is value in the platform
• There are also new instructions and etc that the Xeon D has due to the Broadwell-DE feature set v. the Rangeley core. AVX 2.0, VT-d, TSX-NI, TXT, DPDK and etc.

I still really like the C2558 and C2758 for QAT and am working with Intel to show what that does in a very clear manner.

 

[mackle]

Denverton (when it arrives) needs 10Gbit onboard and a decent (but not earth shattering) bump in performance for me to choose it over the entry Xeon-D's.

I'm looking forward to the power consumption tests, both at idle and under load.