[HN Gopher] New Loongson chip matches Intel's 14600K in IPC test...
___________________________________________________________________
New Loongson chip matches Intel's 14600K in IPC tests, overclocked
to 3 GHz
Author : okasaki
Score : 53 points
Date : 2023-11-29 17:00 UTC (6 hours ago)
(HTM) web link (www.tomshardware.com)
(TXT) w3m dump (www.tomshardware.com)
| tyingq wrote:
| MIPS64 unless Loongson changed direction recently.
| jrockway wrote:
| I remember a lot of noise when Loongson added support for their
| CPUs to Go. Go already had MIPS64, so if they had to do porting
| work, it might not be just a MIPS clone. I haven't looked at
| either architecture in detail for a long time, though.
| stonogo wrote:
| LoongArch is distinct from, but derived from, MIPS64. It's
| starting to look a little RISC-V-like these days.
| MaxBarraclough wrote:
| There's a short section section about this on Wikipedia:
| https://en.wikipedia.org/wiki/Loongson#LoongArch
| Pet_Ant wrote:
| Why the custom arch? If building a CPU is a challenge, then
| supporting an entire toolchain and ecosystem for it is a
| whole other level of complexity that isn't necessary or
| helpful. Moving from being pure MIPS only makes sense if
| there is some patent issue that you are trying to avoid
| because you want to sell to foreign markets to raise volume
| of production.
|
| But then just go with RISC-V, I mean there are legions of
| ravenous fanboys out there _and_ it's already got support.
|
| So the only thing that makes sense is that you have legacy
| MIPS clients and machines where either you want a drop-in
| replacement, or compatibility with them... but also want to
| piggy back on RISC-V and will end up with a CPU with a split
| personality.
| throwbadubadu wrote:
| Why?
|
| "The release shows that domestic CPUs have reached a new
| height in self-sufficiency and performance. It also
| demonstrates that China is capable of producing first-class
| products based on a proprietary CPU structure,"
|
| As this is also kind of one of the responses to the West
| trying to technology-boycot them (what can be considered
| now hugely failed and soon, if not already, backfiring..)
| could also be seen as a feature, and not a problem?! ;)
| dundarious wrote:
| We'll have to wait until China's ASML stockpile is "old"
| tech (in 5 years or so), before deciding if the tech
| embargo has worked. I don't support the embargo, by the
| way.
| fweimer wrote:
| It's possible that they looked at RISC-V and concluded that
| it was unlikely they could achieve their performance goals
| with that architecture.
|
| As far as I understand it, the new Loongson CPU and the
| Asus mainboard are available commercially today. I don't
| think there are any RISC-V CPUs with compare single-core
| performance.
| hgomersall wrote:
| Or they've been at it for a number of years and predated
| risc v.
| pantalaimon wrote:
| If you want to add features to MIPS who would you even talk
| to?
|
| And afair LoongArch started before RISC-V was as popular as
| it is today.
| leecb wrote:
| What operating system would be run by PCs using this chip? Are
| there existing operating systems that support LoongArch?
| kbrackbill wrote:
| Linux
| (https://en.wikipedia.org/wiki/Loongson#Operating_systems)
| matrix12 wrote:
| Hmm no reference to the OpenBSD port.
| https://www.openbsd.org/loongson.html
| pantalaimon wrote:
| This is for the old MIPS based machines, the new ones
| have a custom ISA called LoongArch
| Pet_Ant wrote:
| It is exciting to have more competition in the CPU space. While
| it is currently possible to build a non-x86-64 desktop class
| computer, all the current options are really expensive (looking
| at you Talos). It would be nice to have more architectures at
| this level eventually. Even if only to see what the real costs of
| the 8086-legacy support is to modern performance (if any).
| zamadatix wrote:
| It'll be interesting to see the Snapdragon X Elite next year.
| It won't take the performance crown but it should finally give
| reasonably priced ARM outside of the Apple ecosystem something
| in the high performance class.
|
| As for the real costs the upper bound seems to be no more than
| ~30% on average for the software layers to do it so probably
| less than that if the hardware is built from the ground up to
| run x86 code only. My guess is the real difference is <10%.
| Still appreciable, but not something that really drives the
| gaps.
| Pet_Ant wrote:
| Considering how most of the software we run is open source, I
| don't think legacy support is that important. I think lots of
| organizations would be happy to switch for 10% improvement in
| performance or power.
|
| Open source could really facilitate a golden age in CPU
| design where backwards compatibility really doesn't matter.
| charcircuit wrote:
| Not all open source code is trivial to port to a new
| architecture.
| sangnoir wrote:
| > Considering how most of the software we run is open
| source, I don't think legacy support is that important
|
| At the bottom of the open software pyramid are drivers, and
| they are often closed-source. Qualcomm is a major
| contributor to a lot of Android devices having short
| support periods. Even though Android is open source,
| Qualcomm's drivers are not, and Qualcomm would rather have
| consumers buy devices with the latest snapdragon than keep
| using an upgraded device from 5 years ago.
| Pet_Ant wrote:
| Device drivers aren't really relevant to servers and
| cloud instances which I had in mind where power savings
| and TCO rule.
|
| But yes, for consumer electronics it's a whole other
| story.
| pantalaimon wrote:
| What drivers are you talking about? Most of the divers
| required to run a Snapdragon SoC are in the upstream
| kernel.
| wmf wrote:
| Previous Qualcomm laptops have been very expensive
| unfortunately.
| scq wrote:
| Snapdragon X Elite is faster than all but the M3 Max.
|
| From what I've seen it actually beats the M3 series in
| single-core performance.
|
| That Nuvia acquisition has really paid off for Qualcomm.
| darklion wrote:
| > Snapdragon X Elite is faster than all but the M3 Max. >
| From what I've seen it actually beats the M3 series in
| single-core performance.
|
| Citations? Because from what I have found [1], Qualcomm
| only claims it's 7% faster than a non-Pro, non-Max M2 in
| single-core. It's 50% faster than the M2 in multi-core but
| that's using 50% more cores
|
| And all of this on a system with a TDP almost 4 times that
| of the M2 system, meaning it's consuming a lot more power
| to get essentially the same performance.
|
| All of this on a chip that's coming out in 2024 at the
| earliest, when Apple will be introducing the M4 series --
| two generations down the road from what the Elite X is
| performing at.
|
| --
|
| [1] https://www.tomshardware.com/news/snapdragon-x-elite-
| outperf...
| 2OEH8eoCRo0 wrote:
| Be careful what you wish for. I don't want this kind of
| competition. You'd have to be crazy to use these given the
| sheer amount of tech theft and hybrid warfare coming out of
| China.
| andy99 wrote:
| Does anyone know what kind of compiler support exists or is
| planned for these?
|
| Edit: https://en.wikipedia.org/wiki/Loongson#Compiler_support
| evanjrowley wrote:
| For these next-gen CPUs? No idea.
|
| For Longsoon in general, it was apparently enough support for
| RMS to use one as far back as 2010:
| https://usesthis.com/interviews/richard.stallman/
| doublepg23 wrote:
| I managed to score one and it gives me a new appreciation for
| the word "slow".
| dochtman wrote:
| A bunch of effort has been invested in the Loongson target for
| Rust recently:
|
| https://doc.rust-lang.org/nightly/rustc/platform-support/loo...
| kardianos wrote:
| The Golang loong64 linux build is all "green":
| https://build.golang.org/
| fweimer wrote:
| These seem to have slightly more technical detail:
|
| https://www.tomshardware.com/news/loongson-launches-3a6000-c...
|
| https://wccftech.com/china-loongson-3a6000-cpu-reaches-core-...
| dang wrote:
| Ok, let's use the first one. Changed from
| https://www.ecns.cn/news/sci-tech/2023-11-29/detail-ihcvixpi...
| above. Thanks!
| 2OEH8eoCRo0 wrote:
| Which foundry produces Loongson CPUs?
| scq wrote:
| SMIC (which has been sued multiple times by TSMC for stealing
| their IP and trade secrets)
| huytersd wrote:
| Oh wow so this chip is completely vertically integrated and
| manufactured in China. That's big.
| audunw wrote:
| SMIC uses western manufacturing equipment. Though there's
| efforts on that front too.
| 2OEH8eoCRo0 wrote:
| > Semiconductor Manufacturing International Corporation
| (SMIC) is a partially state-owned publicly listed Chinese
| pure-play semiconductor foundry company.
|
| https://en.wikipedia.org/wiki/Semiconductor_Manufacturing_In.
| ..
|
| Zoinks!
| tambourine_man wrote:
| My guess is that people laughing at how hot and slow this chips
| are now won't in 5-10 years.
| cyanydeez wrote:
| because there'll be a more relevant processor to laugh at?
| olyjohn wrote:
| This is like the Hyundai (automobile) of CPUs. When Hyundai
| started competing in the US, it's cars were slow, cheap and
| kinda crappy. They got some foothold by being cheap above
| all... But now, their cars are formidable competition to some
| of the best out there, and are leading the way in a lot of
| areas. Intel better not get overconfident again, these guys
| could catch up before they know it.
| manuelabeledo wrote:
| Cars are hardly comparable to chips.
| Strom wrote:
| Hyundai didn't have the problem of requiring people to
| build roads differently. Loongson however will have that
| problem, as software is either targeting x86 or ARM.
| blooalien wrote:
| > "Loongson however will have that problem, as software
| is either targeting x86 or ARM."
|
| This is one (of many) advantages of open source software.
| If you have access to the source, you can build it on a
| ridiculously wide range of hardware. I've been downright
| _shocked_ over the years at some of the strange things
| people have run _actually useful_ (and entertaining) open
| source software on. Doom runs pretty much _everywhere_
| just for one example. ;)
| baz00 wrote:
| We'll have newer hot and slow chips by then like we always
| have.
| manuelabeledo wrote:
| Maybe. Maybe not.
|
| I don't think anyone is suggesting that China isn't capable of
| building an efficient and powerful CPU or not, but will they,
| at any point, catch up to _contemporary_ CPUs from Intel, AMD,
| Apple, etc.?
| baz00 wrote:
| I am suggesting that they can't. They have a serious
| innovation problem. A lot of their designs started out stolen
| or cloned from the West. They also don't have anything like
| ASML available to them. And to get this out SMIC ripped off
| TSMC intellectual property.
|
| I'm not saying they won't deliver useful silicon but they
| won't be delivering comparable silicon. And it'll be under
| trade embargo no doubt due to the IP issues so they will have
| a very limited market to scale into.
|
| This happened before (GDR) and the products were never
| competitive.
|
| I think the best thing they have done though is kick out
| compatible ATmega parts and the STM32 parts. They're really
| killing it there.
| meepmorp wrote:
| > A lot of their designs started out stolen or cloned from
| the West
|
| IIRC, Loongson was a straight copy of MIPS with the names
| changed (in the first few iterations, at least).
| baz00 wrote:
| The subtitle says more than the header:
|
| _" Still basically an Intel Core i3-10100 competitor"_
| TillE wrote:
| Sure, but Apple's phone chips started out with performance
| comparable to low-end laptop CPUs.
|
| Looking at the TDP, these seem more promising than other
| Chinese silicon designs which have been terrible flops.
| baz00 wrote:
| Apple have a lot less problems to contend with than China do.
| flopsamjetsam wrote:
| Is there much info about the LoongBT x86 and ARM emulation
| instructions? I can see there's stuff in qemu/hw/loongarch and
| target/loongarch, but would love a more general overview.
___________________________________________________________________
(page generated 2023-11-29 23:01 UTC)