Amazon EC2 M9g Instances(aws.amazon.com)
aws.amazon.com
Amazon EC2 M9g Instances
https://aws.amazon.com/ec2/instance-types/m9g/
82 comments
No benchmarks. No FLOPs. No comparison to commodity hardware. I hate the cloud servers. "9 is faster than 8 which is faster than 7 which is faster than 6, ..., which is faster than 1, which has unknown performance".
You need to benchmark a new EC2 instance anyway. If it’s out of spec, burn it down and redeploy.
Why is that needed, and how would you know if it’s out of spec?
How: You've ran the test on a bunch of hosts and create a spec from ranges.
Why: you might be concerned with network connectivity (you don't get to choose which data center you launch in and it might not be exactly equal), noisy neighbors on shared hosts, etc. if you're measuring for networking, you probably are spinning ups separate accounts/using a bank of accounts and something in every az until you find what you're looking for.
Why: you might be concerned with network connectivity (you don't get to choose which data center you launch in and it might not be exactly equal), noisy neighbors on shared hosts, etc. if you're measuring for networking, you probably are spinning ups separate accounts/using a bank of accounts and something in every az until you find what you're looking for.
Why: you’re paying for a service level that has no guarantee from the vendor
As soon as they're publicly usable people benchmark them carefully. All currently available models have clear metrics.
Since it will be a virtual machine, its performance can be arbitrarily reduced.
If you're interested in using them you should just bench them yourself.
I’ve had terrible luck benchmarking EC2. Measurements are too noisy to be repeatable. The same instance of the wrong type can swing by double digit percentages when tested twice an hour apart.
Who exactly believes manufacturer benchmarks? Just go run your benchmarks yourself and pick. Price/performance is a workload thing.
In Amazon's Graviton 5 PR they note that over half of all new compute capacity added to AWS over the past three years has been Graviton-based. That's an amazing stat.
It really is incredible how ARM basically commoditized processors (in a good way).
It really is incredible how ARM basically commoditized processors (in a good way).
Inversely, I think it's siloed things in somewhat unhealthy ways. We now have a number of vendors that sell/rent you machines that are not generally purchasable. I don't think we've seen too many negative consequences yet, but if things continue in this direction then choosing a cloud provider for a high performance application (eg, something you'll want to compile to machine code and is therefore architecture specific in some way as opposed to a python flask app or something), one may have to make decisions that lock one into a particular cloud vendor. Or at least, it will further increase the cost of changing vendors if you have to significantly tweak your application for some oddities between diff arm implementations at different hosting providers, etc.
I would much rather see some kind of mandatory open market sale of all cpu lines so that in theory you can run graviton procs in rackspace, apple m5 servers in azure, etc.
I would much rather see some kind of mandatory open market sale of all cpu lines so that in theory you can run graviton procs in rackspace, apple m5 servers in azure, etc.
Graviton CPUs are just Neoverse cores (V3 in this case). While it’s true that you can’t just buy a box with the same cores, the cores are basically the same as what you’ll get on a Google or Azure cloud instance (eventually… the latter two have yet to make available anything with Neoverse V3 yet).
Strangely the first-generation Graviton chips have actually shown up in MikroTik hardware that you can just buy. Amazon must be selling off their stock to third parties once it's phased out of use at AWS, but I doubt they'll ever sell the stuff they're still using.
MikroTik is one of the few public organizations who can still get chips from Annapurna Labs after their acquisition by AWS, it seems. Many of their offerings are still using Annapurna parts, and Annapurna still appears as a distinct brand in AWS marketing for its custom silicon, even. I wonder what the specifics of that relationship are.
(Side note but the original Graviton1 was just 16x Cortex-A72 cores, nothing particularly special about it. Actually, all of the Graviton series are just standard ARM cores. But beyond that the SKU they use is indeed the same one AWS uses.)
(Side note but the original Graviton1 was just 16x Cortex-A72 cores, nothing particularly special about it. Actually, all of the Graviton series are just standard ARM cores. But beyond that the SKU they use is indeed the same one AWS uses.)
> over half of all new compute capacity added to AWS over the past three years has been Graviton-based. That's an amazing stat.
Yes and maybe no. They do "cheat" in that internal / managed services often use Graviton where possible. It works out cheaper without the Intel / AMD "tax".
Yes and maybe no. They do "cheat" in that internal / managed services often use Graviton where possible. It works out cheaper without the Intel / AMD "tax".
Pricing when? :(
https://aws.amazon.com/ec2/pricing/on-demand/
https://aws.amazon.com/ec2/pricing/on-demand/
https://instances.vantage.sh/ recently added alerts for any pricing changes on EC2, including newly launching new instances. The site rebuilds every 4 hours so it usually breaks the pricing news first. I have it on for myself and its super helpful just to see when AWS changes things.
[Disclaimer, I'm CEO of Vantage - the company that maintains the site]
[Disclaimer, I'm CEO of Vantage - the company that maintains the site]
Are the built in AWS cost monitoring tools so bad that multiple businesses( including yours) exist just to monitor cost externally ?
Or is your value proposition for companies that use a bunch of different cloud providers ?
Or is your value proposition for companies that use a bunch of different cloud providers ?
[Not a sales pitch - just answering the questions]
This AWS EC2 site is just an open-source project and site we maintain for the benefit of the community. So it's not directly our business but it promotes our brand and is just a helpful site that I think should exist. It's very popular and has been around for about 15 years now.
Our main business hosted on the main domain of https://www.vantage.sh/ is around cloud cost management across 25 different providers (AWS, Azure, Datadog, OpenAI, Anthropic, etc) and the use-cases there about carving up one bill and showing specific costs back to engineers for them to be held accountable to, take action on, etc. Cloud costs and their impact on companies' margins is a big enough problem for vendors like us to exist and we're one player in a larger market.
This AWS EC2 site is just an open-source project and site we maintain for the benefit of the community. So it's not directly our business but it promotes our brand and is just a helpful site that I think should exist. It's very popular and has been around for about 15 years now.
Our main business hosted on the main domain of https://www.vantage.sh/ is around cloud cost management across 25 different providers (AWS, Azure, Datadog, OpenAI, Anthropic, etc) and the use-cases there about carving up one bill and showing specific costs back to engineers for them to be held accountable to, take action on, etc. Cloud costs and their impact on companies' margins is a big enough problem for vendors like us to exist and we're one player in a larger market.
No shame in a sales pitch as long as it's transparent!
I think I see the value here. In any large organization seeing where 100% of your cloud spending is going becomes a time consuming task unto itself.
I think I see the value here. In any large organization seeing where 100% of your cloud spending is going becomes a time consuming task unto itself.
FYI, it's been almost two years since us-east-1-chi-2 launched and it's still missing on the site. Any reason for this? Kind of feels like local zones are a 2nd class citizen even at AWS itself.
Thank you for maintaining this, I do use it every few months at $DAYJOB and it's quite useful for my capacity/deployment planning.
Thank you for maintaining this, I do use it every few months at $DAYJOB and it's quite useful for my capacity/deployment planning.
Huh...weird. I just looked into this and we do have us-east-1-chi-1....but not chi-2. It appears that chi-2 isn't being returned at all as part of the bulk pricing API from AWS for some reason. We'll see if we can dig deeper on why that's the case but just wanted to let you know that's the root issue from our end.
Hey, big fan of the vantage instance finder. Would it be possible to add instance type labels similar to what AWS calls them on their website - "storage-optimized", "general-purpose" etc.?
I find this often useful to quickly compare similar instance types, e.g.: m7g vs. m8g vs. m9g.
I find this often useful to quickly compare similar instance types, e.g.: m7g vs. m8g vs. m9g.
Good feedback! Having someone on our team get this completed for you.
Can you open source the code so I can fix bugs
It's been open-source for 10+ years and the repo is linked in the top right of the site. Here's a link: https://github.com/vantage-sh/ec2instances.info
We'd greatly appreciate contributions!
We'd greatly appreciate contributions!
discussed a couple days ago: https://news.ycombinator.com/item?id=46191993
AWS introduces Graviton5–the company's most powerful and efficient CPU (14 comments)
AWS introduces Graviton5–the company's most powerful and efficient CPU (14 comments)
If only dedicated game servers could run on aarch64...
I've been experimenting FEX on Ampere A1 with x86 game servers but the performance is not that impressed
I've been experimenting FEX on Ampere A1 with x86 game servers but the performance is not that impressed
Doesn't help that Unity requires forking over a pile of cash just to build for Linux ARM ("Embedded Linux") and everything else is free.
General purpose not AI specific? I can't believe it.
AWS has plenty of AI specific offerings for EC2. The P, G and Trn families hit a wide range of AI use cases. Why wouldn't they also offer a general purpose one for typical compute?
Plus with the AI boom, making sure that general purpose compute jobs aren't competing for valuable GPUs is very worthwhile...
You missed the touch of sarcasm. It's a joke, recent AWS announcements have been heavily AI-focused.
I don't really see how this is a productive comment for the article. Most of big tech focuses on AI and those typically get traction in the news. AWS specically has plenty of non-AI announcements: https://aws.amazon.com/new/
Parent comment made a low quality joke that lacked substance.
Parent comment made a low quality joke that lacked substance.
I think that is a joke that reflects pretty well the feeling of many people (me included) that miss the ten years ago AWS and their ability to amaze us with solutions for practical problems, instead of marketing claims on PowerPoints.
Is there a list of Geekbench performance metrics for the various Graviton CPUs?
I need a reference point so I can compare it to Intel/AMD and Apple's ARM cpus.
Otherwise it is buzzwords and superlatives. I need numbers so I can understand.
I need a reference point so I can compare it to Intel/AMD and Apple's ARM cpus.
Otherwise it is buzzwords and superlatives. I need numbers so I can understand.
https://instances.vantage.sh/ shows coremark scores for each EC2 instance type.
It always strikes me that the best place of information for a cloud provider is not from that provider but a third party website. This is not a good comment for the cloud provider.
Funny story: When I was at AWS, I found that the easiest way automate instance data collection was by using the Vantage website code (it's on GitHub).
The cobbler's children have no shoes.
The cobbler's children have no shoes.
Founder of Vantage here and former AWS employee.
We actually recently made the decision to staff someone full time on the site just to maintain it for the community. Even the JSON file for the site gets hit hundreds of thousands of times per day...feels like it's become kind of the de-facto source of truth in the community for where to get reliable AWS pricing information and I believe its powering a pretty remarkable amount of downstream applications with how much usage its getting.
We acquired the site almost 5 years ago and want to continue to improve it for the community. If you have any cloud cost management needs, we're also able to help for our main business here: https://www.vantage.sh/
Awesome to see all the comments on it here!
We actually recently made the decision to staff someone full time on the site just to maintain it for the community. Even the JSON file for the site gets hit hundreds of thousands of times per day...feels like it's become kind of the de-facto source of truth in the community for where to get reliable AWS pricing information and I believe its powering a pretty remarkable amount of downstream applications with how much usage its getting.
We acquired the site almost 5 years ago and want to continue to improve it for the community. If you have any cloud cost management needs, we're also able to help for our main business here: https://www.vantage.sh/
Awesome to see all the comments on it here!
Thry have to adhere to their marketing words and numbers like "efficiency increase of 99999% in performance per dollar per token per watt per U-235 atom used"
Also, use the ffmpeg fps column to check single threaded score.
While the 5 variant isn't yet available outside of the preview, you can of course spin a 4 up and run geekbench yourself. Plenty of people have and you can find them in the GB DB. And of course most people spin up their specific workload to see how it compares.
Core per core it pales compared to Apple's superlative processors, and falls behind AMD as well.
But...that doesn't matter. You buy cloud resources generally for $/perf, and the Graviton's are far and away ahead on that metric.
Core per core it pales compared to Apple's superlative processors, and falls behind AMD as well.
But...that doesn't matter. You buy cloud resources generally for $/perf, and the Graviton's are far and away ahead on that metric.
Not true at all. Single thread CPU scores for Graviton2 are about half that of Intel, while only being about 20% cheaper at best.
Groan. Yes, absolutely true.
While I know this thread will turn into some noisy whack-a-mole bit of nonsense, an easy comparison is the c8g.2xlarge vs the c8i.2xlarge. The former is Graviton 4 vs Granite Rapids in the latter. Otherwise both 16GB, 15Gbps networking, and both are compute optimized, 8 vCPU machines.
Performance is very similar. Indeed, since you herald the ffmpeg result elsewhere the Graviton machine beats the Intel device by 16%.
And the Graviton is 17% cheaper.
Like, this is a ridiculous canard to even go down. Over half of AWS' new machines are Graviton based, but per your rhetoric they're actually uncompetitive. So I guess no one is using them? Wow, silly Amazon.
While I know this thread will turn into some noisy whack-a-mole bit of nonsense, an easy comparison is the c8g.2xlarge vs the c8i.2xlarge. The former is Graviton 4 vs Granite Rapids in the latter. Otherwise both 16GB, 15Gbps networking, and both are compute optimized, 8 vCPU machines.
Performance is very similar. Indeed, since you herald the ffmpeg result elsewhere the Graviton machine beats the Intel device by 16%.
And the Graviton is 17% cheaper.
Like, this is a ridiculous canard to even go down. Over half of AWS' new machines are Graviton based, but per your rhetoric they're actually uncompetitive. So I guess no one is using them? Wow, silly Amazon.
The latter is a 4-core machine with 8 HyperThreads. This doesn't actually matter to your price-performance metric but is worth mentioning because it's the reason why the Intel part performs so comparatively poorly. They're fast chips, they're just wildly uneconomical. If you wanted to compare equal core counts (c8i.4xlarge vs. c8g.2xlarge), then the Intel instance type wins on performance but the Graviton is 58% cheaper.
Groan. Absolutely not. :)
c8g passmark score: 1853 c8i passmark score: 3008
I guess the fps column isn't a good representation of single thread score. Also looking at the passmark scores for i4i vs i4g, i4g is about 1k and intel is about 2k, and the more modern Graviton equivalent of i4 is the same price, so...
https://go.runs-on.com/instances/ec2/c8g
https://go.runs-on.com/instances/ec2/c8i
https://go.runs-on.com/instances/ec2/i4g
https://go.runs-on.com/instances/ec2/i4i
Silly amazon.
c8g passmark score: 1853 c8i passmark score: 3008
I guess the fps column isn't a good representation of single thread score. Also looking at the passmark scores for i4i vs i4g, i4g is about 1k and intel is about 2k, and the more modern Graviton equivalent of i4 is the same price, so...
https://go.runs-on.com/instances/ec2/c8g
https://go.runs-on.com/instances/ec2/c8i
https://go.runs-on.com/instances/ec2/i4g
https://go.runs-on.com/instances/ec2/i4i
Silly amazon.
So confident. And exactly the whack-a-mole nonsense I predicted.
See the comment by electroly. They actually know what they're talking about.
See, the FPS score is for the whole machine. The c8g gives you 8 real cores. The C8i gives you 4 real cores, 4 hyperthreading pseudo-cores. So for those two machines the c8g unequivocally gives you more absolute computing performance, regardless of the passmark single thread (on a single core) on the c8i being better than a single core on the c8g. And the c8g comes at a big discount as well.
That's...the point. The Graviton processors are cheaper per core, and lower performance per core, and you make it up in bulk. You get more performance per $ if you're okay with the ARM stack and your software is good with it, and this is basically universally true comparing Graviton instances versus Intel/AMD alternatives.
You're wrong. Maybe cite some other random nonsense now?
See the comment by electroly. They actually know what they're talking about.
See, the FPS score is for the whole machine. The c8g gives you 8 real cores. The C8i gives you 4 real cores, 4 hyperthreading pseudo-cores. So for those two machines the c8g unequivocally gives you more absolute computing performance, regardless of the passmark single thread (on a single core) on the c8i being better than a single core on the c8g. And the c8g comes at a big discount as well.
That's...the point. The Graviton processors are cheaper per core, and lower performance per core, and you make it up in bulk. You get more performance per $ if you're okay with the ARM stack and your software is good with it, and this is basically universally true comparing Graviton instances versus Intel/AMD alternatives.
You're wrong. Maybe cite some other random nonsense now?
Single thread performance is important for many workloads. It's not nonsense. Things like index builds on an i4g vs i4i could be half as slow. That's really important!
I don't know why you continue to be a fucking asshole. It's just a hosting provider. Go touch grass.
I don't know why you continue to be a fucking asshole. It's just a hosting provider. Go touch grass.
"I don't know why you continue to be a fucking asshole"
Your very first comment was an obnoxious "Not true at all" to the absolutely, incontestably true statement that Gravitons offer better $/perf. So maybe you need to look in the mirror and go touch grass.
Your very first comment was an obnoxious "Not true at all" to the absolutely, incontestably true statement that Gravitons offer better $/perf. So maybe you need to look in the mirror and go touch grass.
Didn't M8g just come out? Am I crazy?
Not crazy. They just have a pretty rapid release cadence for Graviton. New chips ~ every two years.
Happy DEC-10 day, for those who celebrate!
https://www.youtube.com/shorts/Xz-xq-d4jKk
https://www.youtube.com/shorts/Xz-xq-d4jKk
Excited for t5g instances to release... Eventually.
So these are aarch64, right?
More specifically, the CPU cores in AWS Graviton5 are Neoverse V3 cores, which implement the Armv9.2-A ISA specification.
Neoverse V3 is the server version of the Cortex-X4 core which has been used in a large number of smartphones.
The Neoverse V3 and Cortex-X4 cores are very similar in size and performance with the Intel E-cores Skymont and Darkmont (the E-cores of Arrow Lake and of the future Panther Lake).
Intel will launch next year a server CPU with Darkmont cores (Clearwater Forest), which will have cores similar to this AWS Graviton5, but for now Intel only has the Sierra Forest server CPUs with E-cores (belonging to the Xeon 6 series), which use much weaker CPU cores than those of the new Graviton5 (i.e. cores equivalent with the Crestmont E-cores of the old Meteor Lake).
AMD Zen 5 CPUs are significantly better for computationally-intensive workloads, but for general-purpose applications without great computational demands the cores of Graviton5 and also Intel Skymont/Darkmont have greater performance per die area and power consumption, therefore lower cost.
Neoverse V3 is the server version of the Cortex-X4 core which has been used in a large number of smartphones.
The Neoverse V3 and Cortex-X4 cores are very similar in size and performance with the Intel E-cores Skymont and Darkmont (the E-cores of Arrow Lake and of the future Panther Lake).
Intel will launch next year a server CPU with Darkmont cores (Clearwater Forest), which will have cores similar to this AWS Graviton5, but for now Intel only has the Sierra Forest server CPUs with E-cores (belonging to the Xeon 6 series), which use much weaker CPU cores than those of the new Graviton5 (i.e. cores equivalent with the Crestmont E-cores of the old Meteor Lake).
AMD Zen 5 CPUs are significantly better for computationally-intensive workloads, but for general-purpose applications without great computational demands the cores of Graviton5 and also Intel Skymont/Darkmont have greater performance per die area and power consumption, therefore lower cost.
>The Neoverse V3 and Cortex-X4 cores are very similar in size and performance with the Intel E-cores Skymont and Darkmont (the E-cores of Arrow Lake and of the future Panther Lake).
That is not entirely accurate. X4 is big core design. All of its predecessor and successor has always had >1mm2 die space design. X4 is already on the smaller scale, it was the last ARM design before they went all in chasing Apple's A Series IPC. IRRC it was about 1.5mm2 depending on L2 cache. E-Core for Intel has always been below 1mm2. And again IRRC that die size has always been Intel's design guidelines and limits for E-Core design.
More recent X5 / X925 and X6 / X930 / C1 Ultra?? ( I can no longer remember those names ) are double the size of X4. With X930 / C1 Ultra very close to A19 Pro Performance. Within ~5%.
I assume they stick with X4 is simply because it offers best Performance / Die Space, but it is still a 2-3 years old design. On the other hand I am eagerly waiting for Zen 6c with 256 Core. I cant wait to see the Oxide team using Zen 6c, forget about the cloud. 90%+ of companies could fit their IT resources in a few racks.
That is not entirely accurate. X4 is big core design. All of its predecessor and successor has always had >1mm2 die space design. X4 is already on the smaller scale, it was the last ARM design before they went all in chasing Apple's A Series IPC. IRRC it was about 1.5mm2 depending on L2 cache. E-Core for Intel has always been below 1mm2. And again IRRC that die size has always been Intel's design guidelines and limits for E-Core design.
More recent X5 / X925 and X6 / X930 / C1 Ultra?? ( I can no longer remember those names ) are double the size of X4. With X930 / C1 Ultra very close to A19 Pro Performance. Within ~5%.
I assume they stick with X4 is simply because it offers best Performance / Die Space, but it is still a 2-3 years old design. On the other hand I am eagerly waiting for Zen 6c with 256 Core. I cant wait to see the Oxide team using Zen 6c, forget about the cloud. 90%+ of companies could fit their IT resources in a few racks.
Nope. Cortex-X4 is not a big core design, though you are right that at the time of its launch in 2023 the Arm company was not offering bigger cores yet.
The cores designed now by the Arm company for non-embedded applications are distributed into 4 sizes, of which the smaller 2 sizes correspond to what were the original "big and little sizes", but what was originally the big size has been continued into what are now medium-to-small cores, and the last such core before the rebranding was Cortex-A725.
Cortex-X4 is of the second size, medium-to-large. Cortex-X925 was the last big core design before Arm changed the branding this year, so several recent smartphones use Cortex-X925 as the big core, Cortex-X4 as the medium-sized core and Cortex-A725 as the small cores, omitting the smallest Cortex-A520 cores.
Cortex-X4 and Intel Skymont have exactly the same size, 1.7 square millimeter with 1 MB L2 cache memory (in Dimensity 9400 and Lunar Lake). This is about a third of the area of a big core like an Intel P-core and less than a half of the area of a Zen 5 compact core (but AMD uses an older less dense CMOS process; had AMD also used a "3 nm" process the area ratio would not have been so great, and Zen 5 has a double throughput for array operations).
Moreover, Neoverse V3/Cortex-X4 and Intel Skymont/Darkmont have approximately the same number of execution units of each kind in their backends. Only their frontends are very different, which is caused by the different ISAs that must be decoded, Aarch64 vs. x86-64.
The last Arm big core before rebranding, Cortex-X925, was totally unsuitable as a server core, as it had very poor performance per area, having a double area in comparison with Cortex-X4, but a performance greater by only a few tens percent at most. Therefore the performance per socket of a server CPU would have been much lower than that of a Graviton5, had it been implemented with Cortex-X925, due to the much lower number of cores per socket that could have been achieved.
Cortex-X4 was launched in 2023 and it was the big core of the 2024 flagship smartphones, then it has become the medium core of the 2025 flagship smartphones. Its server variant, Neoverse V3, has been launched in 2024 and it has been deployed in products only this year, first by NVIDIA (in Orin) and now by AWS.
It is not at all an obsolete core. As I have said, Intel will have only next year a server CPU with E-cores as good as Cortex-X4. We do not know yet any real numbers about the newly announced Arm cores that have replaced Cortex-A520, Cortex-A725, Cortex-X4 and Cortex-X925, so we do not know if they are really significantly better. The numbers used by Arm in presentations cannot be verified independently and usually when the performance is measured much later in actual products it does not match the optimistic predictions.
The new generation of cores might be measurably better only for computational applications, because they now include matrix execution units, but their first implementation may be not optimal yet, as it happened in the past with the first implementation of SVE, when the new cores had worse energy efficiency than the previous generation (which was corrected by improved implementations later).
The cores designed now by the Arm company for non-embedded applications are distributed into 4 sizes, of which the smaller 2 sizes correspond to what were the original "big and little sizes", but what was originally the big size has been continued into what are now medium-to-small cores, and the last such core before the rebranding was Cortex-A725.
Cortex-X4 is of the second size, medium-to-large. Cortex-X925 was the last big core design before Arm changed the branding this year, so several recent smartphones use Cortex-X925 as the big core, Cortex-X4 as the medium-sized core and Cortex-A725 as the small cores, omitting the smallest Cortex-A520 cores.
Cortex-X4 and Intel Skymont have exactly the same size, 1.7 square millimeter with 1 MB L2 cache memory (in Dimensity 9400 and Lunar Lake). This is about a third of the area of a big core like an Intel P-core and less than a half of the area of a Zen 5 compact core (but AMD uses an older less dense CMOS process; had AMD also used a "3 nm" process the area ratio would not have been so great, and Zen 5 has a double throughput for array operations).
Moreover, Neoverse V3/Cortex-X4 and Intel Skymont/Darkmont have approximately the same number of execution units of each kind in their backends. Only their frontends are very different, which is caused by the different ISAs that must be decoded, Aarch64 vs. x86-64.
The last Arm big core before rebranding, Cortex-X925, was totally unsuitable as a server core, as it had very poor performance per area, having a double area in comparison with Cortex-X4, but a performance greater by only a few tens percent at most. Therefore the performance per socket of a server CPU would have been much lower than that of a Graviton5, had it been implemented with Cortex-X925, due to the much lower number of cores per socket that could have been achieved.
Cortex-X4 was launched in 2023 and it was the big core of the 2024 flagship smartphones, then it has become the medium core of the 2025 flagship smartphones. Its server variant, Neoverse V3, has been launched in 2024 and it has been deployed in products only this year, first by NVIDIA (in Orin) and now by AWS.
It is not at all an obsolete core. As I have said, Intel will have only next year a server CPU with E-cores as good as Cortex-X4. We do not know yet any real numbers about the newly announced Arm cores that have replaced Cortex-A520, Cortex-A725, Cortex-X4 and Cortex-X925, so we do not know if they are really significantly better. The numbers used by Arm in presentations cannot be verified independently and usually when the performance is measured much later in actual products it does not match the optimistic predictions.
The new generation of cores might be measurably better only for computational applications, because they now include matrix execution units, but their first implementation may be not optimal yet, as it happened in the past with the first implementation of SVE, when the new cores had worse energy efficiency than the previous generation (which was corrected by improved implementations later).
Well, also no licensing costs to AMD/intel. So even if at slightly worse performance per chip it’ll end up being cheaper still. AWS doesn’t need to make money on their chips, as they already have the Ec2 margin.
Do you have any insight on when these will be generally available?
Amazon says "Sign up for the preview today".
I have no connection with them, so I have no idea when these instances will be generally available.
Privileged big customers appear to be already testing them.
I have no connection with them, so I have no idea when these instances will be generally available.
Privileged big customers appear to be already testing them.
Yes, Graviton chips are aarch64.
Good question! I read two different Amazon press releases on this but still had to come here for the answer. It seems strange they don't want to advertise the ISA of a compute product - does marketing think it might scare people away?
It seems they don't document the ISA for any instance types. This could be deliberate (and unrelated to marketing) in case they decide to pull features from the instance types in a microcode update. Without any ISA specifics, previous customer commitments towards instance types would still apply.
They list what specific cpus you get for each instance type, see eg https://docs.aws.amazon.com/ec2/latest/instancetypes/gp.html
The part under “Performance specifications“? That's not really a commitment to functionality, especially since there is no vendor specification (ISA reference manual) for many of those model names given. Intel even published a FAQ about missing specifications: https://www.intel.com/content/www/us/en/support/articles/000...
At this point I think they just assume that everyone who cares already know that graviton=arm
>Best price performance
Don't they still offer free nano EC2s? This is not a better price than $0.
Don't they still offer free nano EC2s? This is not a better price than $0.
The free tier for EC2 expires after a year, and the eligible t-family instances have low resources and extreme CPU throttling if you try to do anything more serious with them.
Awhile back I was researching cloud instances for performance, And I noticed that AWS didn't have the latest generations of AMD/Intel. Which are far superior to Graviton 4.
It seems obvious to me that AWS using their market dominance to shift workloads to Graviton.
It seems obvious to me that AWS using their market dominance to shift workloads to Graviton.
I think Graviton would still be much more energy efficient though? (I'm not sure)
I believe the main motivator for AWS is efficiency, not performance. $ of income per watt spend is much better for them on Graviton.
I believe the main motivator for AWS is efficiency, not performance. $ of income per watt spend is much better for them on Graviton.
This sort of makes sense. If there is no competitive advantage in buying the latest AMD or Intel CPUs, why buy them when you can just deploy a generic (ARM licensed) CPU at cheaper prices.
The competitive advantage right now is in NVIDIA chips and I guess AWS needs all their free cash to buy those instead of non-competitive advantage CPUs.
The competitive advantage right now is in NVIDIA chips and I guess AWS needs all their free cash to buy those instead of non-competitive advantage CPUs.
At what point was that true? For example right now ec2 has granite rapids cpus available which are very much the latest and greatest from intel.
I imagine it can take time to actually validate and build out that new infrastructure at scale after AMD/Intel announces these products to the market. It wouldn't surprise me if hyperscalers like AWS, Google, Microsoft, et. al. get a little bit of early previews of this hardware, but it still takes time to negotiate sales, buy the chips, and then actually receive the new chips and make actually useful systems.
Meanwhile, when AWS announces a new chip its probably something they have already been building out in their datacenters.
Meanwhile, when AWS announces a new chip its probably something they have already been building out in their datacenters.