Additionally, the benchmark was only run on one core, and since the last-generation Kunpeng 920 capped out at 64 cores, this result was likely run on a virtual machine or some configuration where only one core was tested. That in turn means the multi-core score isn’t really useful and can’t tell us too much about how performant the full chips are.
Unfortunately, without a look at multi-threaded performance, power consumption, and efficiency, it’s hard to say how competitive Taishan V120 cores will be. For servers in particular, power and efficiency are key due to the cost of electricity, and even if Huawei’s latest server CPUs are fast, that won’t mean much if they consume tons of power.
Huawei is China’s showpiece, they have a lot of motivation to exaggerate. It’s probably not fake, but I do expect this to be a cherry-picked top 1% result.
This is great news though, the more competition, the better.
A cherry-picked top 1% result is sort of what single-core performance is meant to evaluate.
I’m so tired of people trying to ridiculously imply that everything coming out of Huawei is fake. The argument doesn’t even make sense as this isn’t from Huawei, this is an engineering sample given to an end user.
I realize you’re not exactly saying that, but it really gets on my nerves these days. For example, when the Kirin 9000s came out, people were saying Huawei was lying about it being a 7nm chip and that it was made in Taiwan from old stock or whatever. The problem with those claims are the only reason we knew it was a 7nm chip was because a CANADIAN company opened it up and determined it was indeed a 7nm chip. Huawei didn’t say anything and people were calling them liars for what Canadians were saying. It’s why there was the 4.9999G jokes since Huawei never claimed the 9000s was 5G.
Similarly in this case, it’s an engineering sample. Huawei again didn’t say anything. If you’re going to call someone a liar the person you’re accusing should have at least said something.
It seems I misunderstood the situation. The Twitter user in question that uploaded the results seems to specialize in tech leaks. There’s a decent chance that this wasn’t even green lit by Huawei.
Yeah, I know you probably didn’t mean to imply that. To be honest in some ways I was using your post to soapbox. Just tired of people calling Huawei a liar when they literally didn’t say anything.
Oh please, another sinophobe. AMD and NVidia are America’s showpieces as well but you aren’t saying shit about that.
But you’re probably American, so forever and always: CHINA BAD
Fuckin’ idiots.
Oh please, another hapa with an identity crisis
This is good news. I’m looking forward to more competition for chips. This will be better for us consumers. Prices will go down, and quality will go up.
its for servers because its not tied down by consumer software.
Huawei doesnt have a x86-64 license so they wouldnt have access to consumer desktop/laptop software, unless they choose to go arm and design a translation layer, on top of get a major os company onboard to use their design.
one of the strongest positions apple has is they control the entire vertical stack for their business except the fab, so its easier to tightly integrate hardware design to the consumer programs. huaweis software stack wouldnt hold up in its current state. servers dont care because all of their software is tailored specifically to hardware by said company.
Give them a few years, they’ll ace that too
developing an OS isnt that easy. even in apples case, x86 emulation has a lot of penalties for high performance computing.
Linux and RISC-V exist.
and the adoption rate of risc-v and linux combined on consumers is?
im not saying its impossible, but the discussion ia about the viability for consumers, and unless you have a plan on getting a lot consumers on it, its not going to be easy.
if risc-v + linux is already big, Pine64 would be rolling in cash right now because thats their market but clearly it hasnt happened yet.
Linux has a %100 market share in servers. It was around 4% on the Desktop. If the Chinese start adopting Linux desktop en-masse, you can see those numbers change overnight.
China now makes domestic chips for mobile and server. Desktop is not far away.
consumer use of desktop use in china is stringent on pc cafe sales. unless huawei has a plan on having a high performance cpu for existing x86-64 applications, cafes arent going to use them. even leased cpu designs like those by Zhaoxin havent penetrated the market. There have already been several attempts but none has exactly hit yet, because they havent really gone the extra mile for consumer performance. its easy for servers because the company buying the chips tailor their software for the chips. for consumers, its extremely impractical for the hardware company rewrite all software for the hardware.
Bold of you to assume that a Chinese company will care about licenses and patents. When they want to employ this Huawei will either purchase said license on the cheap or they will put up a giant middle finger and disrupt the market and patent holders.
they actually kinda do, and have a history of it…
the Zhaoxin ZX-F series are based on patents that AMD allowed them to use and a joint venture with VIA, who still holds an x86 licence. The Powerstar P3 is essentially a rebranded i3-10105.
companies in China alreay have a history of following up with getting the okay for CPU designs, id argue its bold of you to automatically assume the opposite given history disagrees with you directly in this market.
Huawei does indeed have a good track record and history regarding IP, but I would say there is a cultural expectation of cooperation with IP holders, and if that cooperation is perceived to be lacking then the IP rights are disregarded. The balance of power is not the same as it is with American/European countries. Which is what I mean when I say they will pick up the licensing rights on the cheap, or they will give a middle finger.
The western world is not exactly cooperating with Huawei(for better or worse), and it’s likely that they may “return fire” if given sufficient motivation.
and thats completely bassed on assumptions that they will do it, but with history’s sake, it currently says the opposite, and its kind of haphazard to junp to the extreme end immediately.
the main differemce is that for consumer use, you not only need the hardware, but the cooperation with all the software developers, many which are international to get performamt hardware.
Take for example with GPUs and Moore’s Threads. they can develop a gpu, and repurposing it for server use is doable, as you have he company buying it, programing the software to work with the hardware. but the moment you bring a hardware like that to the consumer space, despite its compute performamce, it fails to launch a lot of typical consumer applications and usecases because of two fronts, consumer software is developers around the world, and you need to support their region or have a large enough market using the hardware for them to even remotely consider optimizing for the hardware, and the drivers need to be well done. Both Moore’s Threads and Intel show that its not remotely easy to make consumer level drivers.
You’re right, I do have a very pessimistic outlook on this subject. I personally see this as an emerging battle of sorts between the technology sector of the West and China, but it may not be that dire. There may still be hope for cooperation and positive competition.
You’re also right about there being a large software development component that would be made less likely by a shaky future. Though I think that could be overcome by force and focus, and I have to admit, that is something I respect about the Chinese tech sector.
I am also pretty amazed in general with the progress that’s been made on these chips in such a short time, props to them for that.
Nice.
Interesting, SMIC 7nm running production and having 5 year old performance instead of 10 year old performance
Call me when Zen 5 launches. Zen 3 is one generation behind AMD’s current flagship.