POWER9 Benchmarks vs. Intel Xeon vs. AMD EPYC Performance On Debian Linux

[Iridium]

Hi Michael, out of curiosity, running the Power9 in BE vs LE gives different results ?
thanks

[chuckula]

Am I the only one around here who feels the iX have lost all meaning and credibility. They can take their i9 and shove it up their ass.

Yes. But if you hate the i9 parts with such a foaming-at-the-mouth-hypocritical-AMD-fanboy passion, then assuming a 7980XE can beat this power PC system at any benchmark whatsoever, are you going to accuse PowerPC of "having lost all meaning and credibility. They can take their POwerPC and shove it up their ass?"

[leiptrstormr]

I checked Talos site. Just under $10k for this configuration.

Yeah, but the Intel and AMD machines cost $20k so $10k is a bargain.

[vegabook]

Not surprising, given that probably millions of man-hours went into x86 compilers' backends and specific code optimization over decades. (multimedia encoders are likely hand-optimized indeed) I for one can't wait to get my hands on one of these Power9 systems.

Doesn't POWER have decades of compiler optimization thanks to massive investment by IBM, Sony, and Apple?

[nuetzel]

We definitely want to see dual AMD EPYC 7601, finally. --- AMD, Tyan??? Bridgman?

[ElectricPrism]

Yes. But if you hate the i9 parts with such a foaming-at-the-mouth-hypocritical-AMD-fanboy passion, then assuming a 7980XE can beat this power PC system at any benchmark whatsoever, are you going to accuse PowerPC of "having lost all meaning and credibility. They can take their POwerPC and shove it up their ass?"

I didn't say anything about benchmarking or PowerPC -- please do not gaslight into topics that I have not commented on filling in the blanks with whatever your brain vomits out in 0.5 seconds.

Intel Marketing has done a good job making noise -- I maintain my opinion that the iX generations are extremely underwhelming.

Now lets talk about my "hypocritical-AMD-fanboy passion" as you tactfully put it -- I think that's a great idea -- perhaps after 15 years of exclusively using Intel on my Desktops and Office I will not renew Intel next upgrade cycle.

You really got an imaginative gen-x evangelical extremist argumentative charm to the tone of your writing. If Satan were a reader and commenter here you two would compete for the crown.

[azdaha]

Wow, I am pissed at myself for sinking 3 grand into this AMD system. Otherwise, I would love to get this Raptor system. FML.

[cybertraveler]

While mentioned why I didn't include price/Watt, I opted to not include price/dollar because it's really tough to get an accurate comparison of the total cost of ownership.... Since these systems tested didn't have identical storage, the Tyan servers had redundant power supplies, different amounts of RAM, etc. So would have been messy trying to compare the total cost of ownership in a fully comparable manner. Especially in the server/workstation market where many customers get discounts from their sales when ordering in any volumes, etc. If just using the CPU price itself, that isn't entirely reflective either with there being many more EPYC/Xeon motherboards available at lower cost than POWER9, so really looking then at just one aspect.

Thanks for the reply.

Yeah, when I wrote my post I did consider how difficult it would be to construct price/watt charts.

Despite what you've said I still think it would have been useful to see basic price info at the top of the article (eg total cost of cpu, mem, mobo and maybe psu for each system). Although the basic info couldn't be used for objective measurements of price/performance, it would provide some readers (or maybe just me ) 1 more useful data point to consider in their mind when looking at the charts.

An example of a datapoint that you did include that I considered when looking at the great benchmarks you provided was that the Talos/POWER system was the only system in the set which has a very open design that requires far less trust of the OEM. If it turned out that the price of all 3 was similar and the average performance of the Talos was only slightly worse, I'd consider the Talos a clear winner for many privacy or security sensitive applications.

I definitely wouldn't expect you to try and calculate total cost of ownership. I doubt even large corporations with big support and accounting teams can figure out that value particularly accurately.

Thanks again.

[pavlerson]

POWER9 cpus are much slower than x86. It is no surprise. It was many years since POWER was faster than x86. Today x86 is much faster and cheaper.

The thing that RISC cpus has an advantage over x86 are mainly two:
1) RAS. RISC cpus such as POWER and SPARC are much more reliable than x86. For instance, you can typically swap everything during run-time, including motherboard, cpu and RAM. Some SPARC cpus can replay instructions if they notice something is wrong. RAS is extremely costly and x86 has no chance to compete in the RAS arena. x86 typically use a cheap cluster, if one fails, just replace it. Quantity before quality.

RISC typically use one large server with 16 cpus or even 32 cpus - and that server must never fail. So everything is extremely good quality and engineered well. Quality before quantity. RAS is very expensive. One single IBM P595 server with 32-cpus costed $35 million. No typo. One single server. You could buy many PCs for $35 million. (It was used for the old TPC-C benchmark - check it up).

2) Scalability. POWER and SPARC scales very well, above 16-cpus. They have scaled to 32 and 64 cpus for decades. x86 does not scale well, and is therefore for small workloads with 4-8 cpus. However, x86 cpus are getting better, so 8 socket x86 servers are capable of handling quite large workloads today. There is not much need for 32-cpu servers today. But business workloads (SAP, databases, etc) are impossible to run on a cluster, so you need a single large server with as many as 16 or 32 cpus. That is where RISC has its place; on extremely large business workloads. The SAP top benchmark all belongs to RISC cpus, they all belong to 32-cpus or so. Not a single x86 server can be seen in the SAP benchmark top. And SAP installations can be extremely expensive and are extremely lucrative, costing more than $200 million. So SAP is a very important benchmark that everybody tries to win. Who does not want to sell one single server for $35 million? Everybody who can, wants to sell such servers. Business servers are extremely profitable. HPC clusters are not, they are just a bunch of cheap PCs on a fast switch. 32 cpu cluster does not cost $35 millions.

BTW. SPARC cpus are many times faster than x86 on a vast array of benchmarks. SPARC M7 and M8 holds many world records.

[Jabberwocky]

Yes. But if you hate the i9 parts with such a foaming-at-the-mouth-hypocritical-AMD-fanboy passion, then assuming a 7980XE can beat this power PC system at any benchmark whatsoever, are you going to accuse PowerPC of "having lost all meaning and credibility. They can take their POwerPC and shove it up their ass?"

How are you trying to be objective with this comment? If you read what soulsource, mastermind, edwaleni, pegasus or pavlerson said (and others that argued with you) then you will notice this benchmark that was requested by community members is not an apples to apples comparison. To me it's more to entertain curiosity rather than to win a pissing contest.