[HN Gopher] Amazon EC2 M9g Instances
       ___________________________________________________________________
        
       Amazon EC2 M9g Instances
        
       Author : AlexClickHouse
       Score  : 142 points
       Date   : 2025-12-06 05:36 UTC (4 days ago)
        
 (HTM) web link (aws.amazon.com)
 (TXT) w3m dump (aws.amazon.com)
        
       | 01HNNWZ0MV43FF wrote:
       | So these are aarch64, right?
        
         | watermelon0 wrote:
         | Yes, Graviton chips are aarch64.
        
         | adrian_b wrote:
         | 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.
        
           | sahilagarwal wrote:
           | Do you have any insight on when these will be generally
           | available?
        
             | adrian_b wrote:
             | 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.
        
           | bushbaba wrote:
           | 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.
        
           | ksec wrote:
           | >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.
        
             | adrian_b wrote:
             | 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).
        
         | dtf wrote:
         | 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?
        
           | fweimer wrote:
           | 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.
        
             | zokier wrote:
             | They list what specific cpus you get for each instance
             | type, see eg https://docs.aws.amazon.com/ec2/latest/instanc
             | etypes/gp.html
        
               | fweimer wrote:
               | 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...
        
           | zokier wrote:
           | At this point I think they just assume that everyone who
           | cares already know that graviton=arm
        
       | stevefan1999 wrote:
       | 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
        
         | Rohansi wrote:
         | 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.
        
       | thewisenerd wrote:
       | 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)
        
       | HatchedLake721 wrote:
       | Pricing when? :(
       | 
       | https://aws.amazon.com/ec2/pricing/on-demand/
        
         | StratusBen wrote:
         | 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]
        
           | 999900000999 wrote:
           | 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 ?
        
             | StratusBen wrote:
             | [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.
        
               | 999900000999 wrote:
               | 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.
        
           | jolan wrote:
           | 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.
        
       | yonisto wrote:
       | 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.
        
         | xeornet wrote:
         | 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.
        
         | zokier wrote:
         | 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.
        
         | jcims wrote:
         | >Which are far superior to Graviton 4.
         | 
         | Not if you are looking at price/performance. AWS could be
         | taking a loss to elevate the product though, no way to know for
         | sure.
        
           | coredog64 wrote:
           | If they were taking a loss, they wouldn't run a crapton of
           | internal workloads on Graviton.
        
         | bhouston wrote:
         | 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.
        
         | vel0city wrote:
         | 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.
        
       | Artoooooor wrote:
       | General purpose not AI specific? I can't believe it.
        
         | jackling wrote:
         | 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?
        
           | swiftcoder wrote:
           | Plus with the AI boom, making sure that general purpose
           | compute jobs aren't competing for valuable GPUs is very
           | worthwhile...
        
             | jackling wrote:
             | Where do you see GPUs in this release? This is a CPU-based
             | instance.
        
               | jldugger wrote:
               | Thats the point -- if you only ship GPU instances then
               | every workload ties up precious GPU time.
        
               | jackling wrote:
               | Yea I misread the parent comment, my bad.
        
           | arrowleaf wrote:
           | You missed the touch of sarcasm. It's a joke, recent AWS
           | announcements have been heavily AI-focused.
        
             | jackling wrote:
             | 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.
        
               | ciberado wrote:
               | 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.
        
       | jit_hacker wrote:
       | Didn't M8g just come out? Am I crazy?
        
         | jcims wrote:
         | Not crazy. They just have a pretty rapid release cadence for
         | Graviton. New chips ~ every two years.
        
       | bhouston wrote:
       | 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.
        
         | sciurus wrote:
         | https://instances.vantage.sh/ shows coremark scores for each
         | EC2 instance type.
        
           | dylan604 wrote:
           | 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.
        
             | coredog64 wrote:
             | 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.
        
               | StratusBen wrote:
               | 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!
        
             | LtdJorge wrote:
             | 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"
        
           | winrid wrote:
           | Also, use the ffmpeg fps column to check single threaded
           | score.
        
         | llm_nerd wrote:
         | 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.
        
           | winrid wrote:
           | Not true at all. Single thread CPU scores for Graviton2 are
           | about half that of Intel, while only being about 20% cheaper
           | at best.
        
             | llm_nerd wrote:
             | 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.
        
               | electroly wrote:
               | 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.
        
               | winrid wrote:
               | 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.
        
               | llm_nerd wrote:
               | 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?
        
             | wmf wrote:
             | Do you realize we're talking about Graviton5 now?
        
               | winrid wrote:
               | The Graviton5 instances I've been comparing to intel are
               | the same price...
        
       | llm_nerd wrote:
       | 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).
        
         | Octoth0rpe wrote:
         | 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.
        
           | sgerenser wrote:
           | 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).
        
             | jsheard wrote:
             | 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.
        
               | aseipp wrote:
               | 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.)
        
         | re-thc wrote:
         | > 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".
        
       | 1970-01-01 wrote:
       | >Best price performance
       | 
       | Don't they still offer free nano EC2s? This is not a better price
       | than $0.
        
         | Kwpolska wrote:
         | 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.
        
       | diath wrote:
       | 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".
        
         | oofbey wrote:
         | As soon as they're publicly usable people benchmark them
         | carefully. All currently available models have clear metrics.
        
         | mmontagna9 wrote:
         | If you're interested in using them you should just bench them
         | yourself.
        
         | itomato wrote:
         | You need to benchmark a new EC2 instance anyway. If it's out of
         | spec, burn it down and redeploy.
        
           | Kwpolska wrote:
           | Why is that needed, and how would you know if it's out of
           | spec?
        
         | renewiltord wrote:
         | Who exactly believes manufacturer benchmarks? Just go run your
         | benchmarks yourself and pick. Price/performance is a workload
         | thing.
        
         | sebazzz wrote:
         | Since it will be a virtual machine, its performance can be
         | arbitrarily reduced.
        
       | DonHopkins wrote:
       | Happy DEC-10 day, for those who celebrate!
       | 
       | https://www.youtube.com/shorts/Xz-xq-d4jKk
        
       ___________________________________________________________________
       (page generated 2025-12-10 23:01 UTC)