New High Performance, x86 Compatible Microprocessors from Centaur / VIA

[lexluthermiester]

You're arguing about mobile CPU's/SOC's in a thread about x86 and desktop processors

That is because some mobile devices contain X86 CPU's. Laptop's contain X86 CPU's and are not modular like PC's. The statement made above by Candle_86 is that the industry has been a "two horse race since 1999". This has never been true, especially today. Additionally, there are many ARM CPU's that are considered very high performance and are currently being used in desktop applications. Your perspective is too narrow.

 

[candle_86]

That is because some mobile devices contain X86 CPU's. Laptop's contain X86 CPU's and are not modular like PC's. The statement made above by Candle_86 is that the industry has been a "two horse race since 1999". This has never been true, especially today. Additionally, there are many ARM CPU's that are considered very high performance and are currently being used in desktop applications. Your perspective is too narrow.

Arm is worthless for servers and Enterprise except on phones. To many custom in house programs written by guys that are no longer with companies, to much legacy software or even current software that isn't compatabile. Arm can't emulate x86_64 period and is slow at x86, I wouldn't want to had people desktops for accounting running on arm, when they already are needing xeon precisions.

It's the reason the iPad pro failed, the software isn't full featured and the compute to slow for real work.

 

[R-T-B]

Meanwhile, ARM64 has deprecated ARM for almost a half decade now and you two are still talking about ARM like it's limitations are relevant.

 

[Durvelle27]

The Geekbench already has a 2GHz sample of the Chinese x86 Zhaoxin KX-7000 processor. 80% IPC Zen 2

source: https://translate.google.com/translate?hl=sk&sl=cs&tl=en&u=https://www.cnews.cz/procesor-via-zhaoxin-kx-7000-unik-geekbench-2ghz/


AMD FX-8300 3300MHz @ 2.00GHz
Single-Core Score: 306 (65,25%)
Multi-Core Score: 1 705 (52,24%)


vs


KX-7000 ZX-F OctaCore 2000MHz @ 2.00GHz
Single-Core Score: 469 (+53,27%)
Multi-Core Score: 3 264 (+91,44%)


source: https://browser.geekbench.com/v5/cpu/compare/526995?baseline=562975


___________________

source: https://browser.geekbench.com/v5/cpu/562975.gb5
source: https://browser.geekbench.com/v5/cpu/562975

P.S. If you have Geekbench 5 results your modern CPU @ 2.00GHz so pls write here! Thank you.

Where did you get 80% Zen IPC when testing against a old crappy Vishera processor which is know to have terrible IPC

 

[candle_86]

Meanwhile, ARM64 has deprecated ARM for almost a half decade now and you two are still talking about ARM like it's limitations are relevant.

It is still relevant, your forgetting x86 and it's extensions are already running on risc using a cisc frontend to Port the code to microinstructions. When arm is as fast as current CPUs it will also be as large and complex, while sucking just as much power negating every advantage while still having the disadvantage of emulating x86 code at a 30-50% penalty. They can't emulate in hardware because they don't have a license so software emulation is required which will always be a big hit.

The only companies that could emulate x86 in hardware are AMD, Intel, and via and none of them are interested.

 

[R-T-B]

When arm is as fast as current CPUs it will also be as large and complex, while sucking just as much power negating every advantage while still having the disadvantage of emulating x86 code at a 30-50% penalty.

Why in the world would a server environment bother with emulation?

They usually run linux or open source software and they'd just recompile to target arm.

No, emulation is never going to replace native x86 in the desktop, but I mean... duh. No one said that. That's why compile targets exist.

 

[candle_86]

Why in the world would a server environment bother with emulation?

They usually run linux or open source software and they'd just recompile to target arm.

No, emulation is never going to replace native x86 in the desktop, but I mean... duh. No one said that. That's why compile targets exist.

but again legacy programs that you no longer have the original creator for or necessary software that is no longer updated because that company is out of bussiness. Meaning you either emulate it on ARM or run it on real hardware. ARM will never gain more than a footnote in server or desktop spaces, its best suited to lower power devices like phones and tablets, leave real work to X86

 

[biffzinker]

The only companies that could emulate x86 in hardware are AMD, Intel, and via and none of them are interested.

Transmeta did a form of emulation with their Crusoe processor through their software layer called code morphing.

Code Morphing Software consisted of an interpreter, a runtime system and a dynamic binary translator. x86 instructions were first interpreted one instruction at a time and profiled, then depending upon the frequency of execution of a code block, CMS would progressively generate more optimized translations

Source: https://en.wikipedia.org/wiki/Transmeta

ARM will never gain more than a footnote in server or desktop spaces, its best suited to lower power devices like phones and tablets, leave real work to X86

ARM could replace X86 as time moves on, and no one (the vast majority) would care or even know. Your way to quick to write off ARM as incapable of competing with X86.

 

[seronx]

Where did you get 80% Zen IPC when testing against a old crappy Vishera processor which is know to have terrible IPC

Gigabyte Technology Co., Ltd. AX370-Gaming K7 vs Shanghai Zhaoxin Semiconductor Co., Ltd. CHA001 MB - Geekbench Browser

CatMerc's AMD's Zen2 8c/16t @ 2 GHz
vs
Zhaoxin's ZX-F 8c/8t @ 2 GHz

Single-threaded => 77% of Zen2 @ 2 GHz
Multi-threaded => 60% of 3700X @ 2 GHz

Should be noted that the ZX-F is an APU-equivelent. So, it's closest competitor would be the Ryzen 7 4700u(8c/8t/7CU).

 

[notb]

Arm is worthless for servers and Enterprise except on phones.

Your main mistake is probably thinking about servers as general-use machines - kind of like large PCs.
So maybe you'd like to run some Windows VMs on them or whatever.

In reality though, large servers are usually run as purpose-built machines - designed for a particular task or system (file storage, database, ERP, engineering computations, networking...).
So it doesn't really matter that all software in the world doesn't work on ARM. It's really enough that the one application you need does.

And in micro servers (because those tiny machines with 2-core Pentiums also matter) ARM's are perfectly sufficient.

Also, don't underestimate popularity of ARM (both in available software and among programmers).
Think about RPi and similar platforms. For people learning coding today (especially for science or engineering) ARM is a very natural choice. They don't feel limited.

 

[R-T-B]

but again legacy programs that you no longer have the original creator for or necessary software that is no longer updated because that company is out of bussiness. Meaning you either emulate it on ARM or run it on real hardware.

This is a very strange use case and usually will not require a lot of horsepower.

ARM could replace X86 as time moves on, and no one (the vast majority) would care or even know.

Yep. This is exactly how Apple killed PowerPC.

 

[notb]

Yep. This is exactly how Apple killed PowerPC.

Very different use case.
Mac owners are generally very feature-oriented, rational. They expect certain things to work out of box (and they're willing to pay).
So as long as Macs work and are easy to work with, Apple customers don't really give a f... what's inside.

PC owners (especially on desktops) are more performance-oriented. They care about benchmarks, about parts inside.
And, sadly, there isn't a "governing body" that guarantees software compatibility (or quality, looks...).

Moving from x86 to another architecture on PCs would be a huge pain.

On servers, at least those deployed as full solutions (where a provider gives you software and hardware) - it happens all the time (between mainframe and x86, ARM will join)

 

[R-T-B]

Very different use case.

I know. My point was more if manufacturers wanted to force arm on consumers they could.

 

[lexluthermiester]

Your way to quick to write off ARM as incapable of competing with X86.

I don't think that is what RTB is they are saying, mostly because it already is.

 

[dorsetknob]

Intel does not compete in the Ultra low Power stakes ....That is the ARM Strong point.
Intel has Tried but soon exited those markets.

 

[R-T-B]

Intel does not compete in the Ultra low Power stakes ....That is the ARM Strong point.
Intel has Tried but soon exited those markets.

X86 tends to have a hard time there, but I'm sure it could be done with enough effort. At that point though the core instructions would be so stripped down you may as well just choose something else.

 

[biffzinker]

I don't think that is what RTB is saying, mostly because it already is.

That was a reply to @candle_86 NOT @R-T-B

 

[Assimilator]

It's pretty obvious (and has been for years) that Apple hopes to transition its entire ecosystem, including desktop, to ARM. It's a viable strategy because most Mac users just want expensive shiny things with a particular brand name stamped on, so they won't even notice ARM's lower IPC. If ARM ever makes it further than mobile phones and ultraportables, it will be due to Apple, not Qualcomm or Amazon or all these weird Chinese companies trying to make ARM work in the server space.

 

[notb]

Intel does not compete in the Ultra low Power stakes ....That is the ARM Strong point.
Intel has Tried but soon exited those markets.

I wouldn't really say Intel exited low power market. They have quite a few ~5W SoCs.
Yes, they aren't investing in the ~2W range, but not because of performance (Atoms were quite good), but because of production cost. There weren't competitive against cheap ARM.

Anyway, we should call Intel quite resilient. AMD stopped developing Geode around 2008.

It's pretty obvious (and has been for years) that Apple hopes to transition its entire ecosystem, including desktop, to ARM. It's a viable strategy because most Mac users just want expensive shiny things with a particular brand name stamped on

I'm not sure where your opinion about Apple customers comes from. I assure you it's incorrect.
Apple users want performance as much as everyone. It's just that in their case this performance has a purpose, i.e. they focus on stuff other than running benchmarks.

We don't have ARM CPUs that could replace Intel's high-end mobile lineup - let alone Xeons in Mac Pro.
If such CPUs ever appear, Apple may try to move.
But given how much x86 gave Apple in the way of software availability, IMO they'll just go for custom x86 CPUs. They can pay Intel or AMD to make them. Or they can just buy AMD.

 

[R-T-B]

AMD stopped developing Geode around 2008.

I mean, did they develop it at all, really? Or just acquire and murder it?

Or they can just buy AMD.

Pretty sure the base x86 license does not survive a buyout.

Of course, Apply may not even NEED x86 support.

 

[lexluthermiester]

That was a reply to @candle_86 NOT @R-T-B

Oops, my bad. I was agreeing with you and even expanding on the idea. ARM is already competing and even out pacing X86 is some area's.

 

[Vya Domus]

ARM could replace X86 as time moves on, and no one (the vast majority) would care or even know. Your way to quick to write off ARM as incapable of competing with X86.

Everyone sort of misses the point, ARM itself is just a collection of licenses to a variety of architectures. ARM doesn't compete directly with anyone, the companies that use those licenses are the ones that could compete with x86 licensed products.

The concept of "ARM vs x86" doesn't really make sense in the real world because that's not really what matters. Here's the thing, ARM do not have dedicated designs made directly for HPC/desktop applications, Intel and AMD do, that's a fact. So no company can just pick up their designs and begin competing with AMD or Intel, it's simply not possible. What they have to do, is come up with their own designs, at which point "ARM" is essentially reduced to just an ISA and the effort of coming up with something competitive falls on said company. That endeavour is colossal without ARM's direct interest in developing architectures for those purposes because you need to construct your product from scratch.

But ARM doesn't even have an interest in competing with x86, they have an interest in selling as many licenses as possible and it turns out that mobile is where they can do that most effectively. Unless they ever change their mind, we're likely never going see ARM competing directly with x86 products in any meaningful manner. Huawei, Amazon and a few other are trying to do that but long term I don't think it's going to go anywhere because it's not a coherent effort.

 

[E-Bear]

The last Via cpu I had was a 1ghz soldered on a pc-chips motherboard.

 

[trparky]

Atoms were quite good

What are you talking about? The Atom chips were garbage in the performance department.

 

[lexluthermiester]

What are you talking about? The Atom chips were garbage in the performance department.

When compared to mainstream desktop CPU's sure, but for what they were intended to be, they performed very well. I still have my ASUS EEEPC with an Atom N330 and NVidia Ion and it still performs well.