Live data from Hacker News

The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

anandtech.com

221–230 of 266 posts

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#221

Earlier quoted context omitted.

> But for that bandwidth to be used efficiently, the processes on each NUMA node need to almost exclusively limit themselves to memory attached to their node -- at which point, well-written software could probably do just as good spread out over several machines that are connected by multiple 100Gb network links, and then you saved two or three bajillion dollars. Wait, so latency over NUMA is too much, but you're wil…

> Wait, so latency over NUMA is too much, but you're willing to incur a 100Gb network link? The point is not just the latency, but latency and bandwidth. If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the RAM bandwidth. It's a total all around bust. You're just wasting money at that point. ^1 If you aren't re…

> If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the memory bandwidth. It's a total all around bust. You're just wasting money at that point.

If the interconnect is your bottleneck, you spend money on the interconnect to make it faster. Basic engineering: you attack the bottleneck.

You don't start talking about slower systems and how they're cheaper. Because that just slows down the rest of your system.

------

If you just wanted cores, you buy a 28core Xeon Gold. The point of 28-core Xeon Platinum is for the 8-way interconnect and scaling up to 8-way NUMA systems. The only person who would ever buy a Xeon Platinum is someone who wants 40GBps UPI connections at relatively low latencies. (or maybe even the 300GBps connections that IBM offers, but that's a similar high-cost vertical-scaling system)

> So you really have to be in a very obscure situation which can't fit onto a dual socket Rome server, but can fit within a machine less than 2x larger. (28 cores * 8 sockets is less than 2x larger than 64 cores * 2 sockets)

That's not even that hard to figure out! A 48TB Memcached / Redis, which is far more useful than a 8TB Memcached / Redis box.

A bit basic, but yeah. That's the point: spend more money on hardware and then don't spend much engineering time thinking about optimization. If 48TBs of RAM solves a problem that 8TB cannot, then just get the 48TB system.

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#222
post #81

Earlier quoted context omitted.

Right? "This platform does/does not run my software" is a huge checkbox when buying computers. The viewpoint of people who actually specify, buy, and operate servers at scale is severely underrepresented in these discussions.

So you're never going to consider moving to another architecture because your software (currently) doesn't run on it? So Amazon made a huge mistake with Graviton then? Last time I checked Amazon 'specify, buy, and operate servers at scale'.

Like many developers, I will be trialing my software stack on Graviton2 instances at AWS next year, in hopes of saving 20-40% costs.

The stack should all work on ARM: Ubuntu (check), Java (check), MySQL (check), Grafana (check), Kafka (check).

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#223

Earlier quoted context omitted.

> I had an Atom tablet [...] and it really wasn't a great device - ran hot and very short battery life. In all honesty the implementation and OS also matter. Most Atom tablets weren't particularly great (or even good) devices overall. And running Windows on a contemporary ARM SoC wouldn't have resulted in a better experience.

It was Android and much, much worse than contemporary Arm tablets.

Most Atom tablets I've ever seen were very bad implementations from noname brands, I haven't seen a halo product. And the OS choice was worse, either a full Windows, or an x86 Android, neither of which is the best choice for such a device. And that's on top of any weakness the CPU itself may have.

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#224

Earlier quoted context omitted.

> Wait, so latency over NUMA is too much, but you're willing to incur a 100Gb network link? The point is not just the latency, but latency and bandwidth. If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the RAM bandwidth. It's a total all around bust. You're just wasting money at that point. ^1 If you aren't re…

> If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the memory bandwidth. It's a total all around bust. You're just wasting money at that point. If the interconnect is your bottleneck, you spend money on the interconnect to make it faster. Basic engineering: you attack the bottleneck. You don't start talking abo…

Preface: If you have actually encountered applications that must be run on 8-socket systems because those are literally the only fit for the application... I would love to hear about those experiences. With the advent of Epyc Rome, most use cases for these 8-socket systems vanished instantly. It would be fascinating to hear about use cases that still exist. Your experiences are obviously different than mine.

If you need more than 8TB of RAM, with the right application design you can probably do better with fast Optane Persistent Memory or Optane SSDs, and an effective caching strategy. You can have many dozens of terabytes of Optane storage connected to a single system, and Optane is consistently low latency (though not as low latency as RAM, obviously).

If you need more compute power, you can generally do better with multiple linked machines. You can only scale an 8-socket system up to 8 sockets. You can link way more machines than that together to get more CPU performance than any 8 socket system could dream of.

----------

I didn't expect you to read and respond so quickly, so I had edited my previous comment before you submitted your reply.

This was a key quote added to my previous comment:

>> So you really have to be in a very obscure situation which can't fit onto a dual socket Rome server, but can fit within a machine less than 2x larger. (28 cores * 8 sockets is less than 2x larger than 64 cores * 2 sockets)

In response to your current comment,

> If the interconnect is your bottleneck, you spend money on the interconnect to make it faster. Basic engineering: you attack the bottleneck.

Exactly. Using a dual-socket Epyc Rome system would be more than half as powerful as the biggest 8-socket Intel systems, but it would reduce contention over the interconnect dramatically, which means that many applications that are simply wasting money on an 8-socket system would suddenly work better.

This also goes back to my comment about using accelerators instead of an 8-socket system.

The odds of encountering a situation that just happens to work well with Intel's ridiculously complicated 8-socket NUMA interconnect, but can't work well over a network, and can't work well on a system half the size and requires enormous amounts of RAM to keep the cores fed, the odds seem vanishingly small... and in that case, we still have to consider whether an accelerator (GPU, FPGA, or ASIC) could be used to make a solution that is a better fit for the application anyways, and if so, you'll save large amount of money that way as well.

So, to make buying an 8-socket system make sense, the application must require performance that is...

- less than twice a dual socket Epyc Rome system, but greater than one dual socket Epyc Rome system can handle

- not dependent on transferring huge amounts of data around the interconnect

- dependent on very low latency communication between NUMA nodes

- needs enormous memory bandwidth for each NUMA node

- needs huge amounts of RAM on each memory channel (so you can't just use HBM2 on a GPU to get massive amounts of bandwidth, for example)

- etc.

It's a niche within a niche within a niche.

As I said in an earlier comment, I probably should be more impressed instead of being so cynical about the usefulness of such a machine. They are engineering marvels... but in almost every case, you can save money with a different approach and get equal or better results.

That's why 8-socket server sales made up such a small percentage of the market, even before Epyc Rome came in and completely obliterated almost all of the very little value proposition that remained.

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#225

Earlier quoted context omitted.

How are Exynos chromebooks terrible?

They're slow. The only thing a Chromebook has is a web browser. The browser performance of e.g. the Samsung Chromebook 2 was less than half that of its contemporary competitors, and a modern Chromebook with an Intel Core CPU is 5-10x faster. Exynos didn't even manage to be the fastest ARM CPU in a Chromebook. The Tegra K1 was a bit quicker.

> The only thing a Chromebook has is a web browser

My (Celeron) Chromebook 2 supported a whole Linux VM (via Crostini) and using Vim for python and Go development on it was noticeably slower than any contemporary laptop - I can't imagine the Exynos version was far off on perf (since it was only a diff SKU). I would not call it "terrible".

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#226

Earlier quoted context omitted.

That’s useful if you’re not averse to MacOS. Given that Apple now deploys backdoors by default for their own apps which will always inevitably result in exploits, and turning their back on decades of computing history with no “legacy” Heck, my 2015 MBP is still running Mojave (and thankfully still receiving software updates)

> Given that Apple now deploys backdoors by default for their own apps which will always inevitably result in exploits You're talking about the certificate revocation check that bypasses VPNs? That one is to prevent malware from being able to hijack and block certificate checks.

No, I’m talking about regular Apple apps that bypass local software firewalls.

See this HN thread from ~2 months ago for an introduction: https://news.ycombinator.com/item?id=24838816

This is new behaviour to Big Sur, breaking existing tooling and providing a back door for Apple, and anyone who can exploit it like this guy https://twitter.com/patrickwardle/status/1327726496203476992

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#227
post #2

These things look awesome, but are pretty expensive, still. I'd love to put an ARM server in my homelab, but the price range between 100$ (RPi & Co) and 10000$ (enterprise server) seems pretty empty right now :/

I'll believe these are a real viable thing for ordinary open source Linux/BSD developers, when I can go buy a $150 motherboard and a $250 CPU from newegg, that have performance anywhere NEAR what I can do with the equivalent priced Ryzen. Or even some Intel 10th/11th generation core-whatever i5/i7 CPU.

Good point, I've been looking at NewEgg and Amazon retail ever since the M1, because I need Ubuntu, not Apple.

My preference is for a mini form factor without a fan that I can run headless in the office lab.

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#228

Earlier quoted context omitted.

> If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the memory bandwidth. It's a total all around bust. You're just wasting money at that point. If the interconnect is your bottleneck, you spend money on the interconnect to make it faster. Basic engineering: you attack the bottleneck. You don't start talking abo…

Preface: If you have actually encountered applications that must be run on 8-socket systems because those are literally the only fit for the application... I would love to hear about those experiences. With the advent of Epyc Rome, most use cases for these 8-socket systems vanished instantly. It would be fascinating to hear about use cases that still exist. Your experiences are obviously different than mine. If you n…

I mean, I buy AMD Threadripper for my home use and experimentation. I'm pretty aware of the benefits of AMD's architecture.

But I also know that in-memory databases are a thing. Nothing I've touched personally needs an in-memory database, but its a real solution to a real problem. A niche for sure, but a niche that's pretty common actually.

Whenever I see these absurd 8-socket designs with 48TBs of RAM, I instinctively think "Oh yeah, for those in-memory database peeps". I never needed it personally, but its not that hard to imagine why 48TBs of RAM beats out any other architecture (including Optane or Flash).

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#229

Earlier quoted context omitted.

They're slow. The only thing a Chromebook has is a web browser. The browser performance of e.g. the Samsung Chromebook 2 was less than half that of its contemporary competitors, and a modern Chromebook with an Intel Core CPU is 5-10x faster. Exynos didn't even manage to be the fastest ARM CPU in a Chromebook. The Tegra K1 was a bit quicker.

> The only thing a Chromebook has is a web browser My (Celeron) Chromebook 2 supported a whole Linux VM (via Crostini) and using Vim for python and Go development on it was noticeably slower than any contemporary laptop - I can't imagine the Exynos version was far off on perf (since it was only a diff SKU). I would not call it "terrible".

The Celeron Chromebooks 2 were only slightly faster, maybe 15%. That's why I specifically called out the Intel Core CPU, not the Celeron. The Celeron N2840 is a "Bay Trail" Atom core, which nobody would willingly purchase. The state of the art CPU from that generation of Chromebooks was the 4th gen Core i3, a "Haswell" part that was more than twice as quick as that Celeron.

Re: The Ampere Altra Review: 2x 80 Cores Arm Server Performance Monster

#230

Earlier quoted context omitted.

> Wait, so latency over NUMA is too much, but you're willing to incur a 100Gb network link? The point is not just the latency, but latency and bandwidth. If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the RAM bandwidth. It's a total all around bust. You're just wasting money at that point. ^1 If you aren't re…

> If the application is relying heavily on the NUMA interconnect to transfer tons of data, it's not going to be making efficient use of the processor cores or the memory bandwidth. It's a total all around bust. You're just wasting money at that point. If the interconnect is your bottleneck, you spend money on the interconnect to make it faster. Basic engineering: you attack the bottleneck. You don't start talking abo…

You have since edited your comment, so...

> That's not even that hard to figure out! A 48TB Memcached / Redis, which is far more useful than a 8TB Memcached / Redis box.

No... dozens of terabytes of Optane would be just as good and much much cheaper. The person designing the system would have to prove that a few nanoseconds of latency difference makes any material difference to the company's profits in order to justify the more expensive machine. Otherwise, it's a huge waste of company money, hurting the business.

Also keep in mind that Redis is only going to be using a single core of that machine. A total waste of huge amounts of CPU cores just to have a lot of RAM, when there are equally good solutions that cost much less.

You surely must see why I'm skeptical.

> That's the point: spend more money on hardware and then don't spend much engineering time thinking about optimization.

It's bad business practice to buy the most expensive thing possible instead of engineering a solution that is the right price. If the more expensive solution saves money in the long term, sure... but your example doesn't show this.

Post reply on HN