Editorial x86 Lacks Innovation, Arm is Catching up. Enough to Replace the Giant?

[efikkan]

The x86 architecture is massive, having more than a thousand instructions, some of which are very complex. This approach is called Complex Instruction Set Computing (CISC). Internally, these instructions are split into micro-ops, which further complicates processor design.

Actually, you got it all wrong.
Back in the 80s there were two main arguments in the CISC vs. RISC discussions;
- A smaller instruction set allows higher flexibility in implementation
- Having to support legacy instructions in hardware
Moving to using micro-operations solved both of these, and therefore eliminating the only real advantages of RISC over CISC. In all x86 CPUs since the early 90s, the CPU front-end decodes the x86 ISA into the microarchitecture's specific ISA, which gives the designer full control over which instructions are prioritized in the hardware implementation, and which can be "simulated" using a combination of other instructions. This way the instructions can be optimized for however many execution units and various resources are present in the CPU, and the whole pipeline doesn't need to support every legacy feature. Every modern x86 microarchtecture is sort of a hybrid of CISC and RISC, with RISC-like micro-operations. Even current ARM microarchitectures have gotten some CISC-like features including some SIMD and loads of ASIC features, but still differs from modern "CISC" designs by using a load-store architectures (like FordGT90Concept mentioned), which means that ARM will always require more instructions to do the same work.

x86 is still advancing, and if anything x86 is more held back by software than ARM is respectively. Very little software is compiled to use any ISA features beyond AMD64/SSE2, even as recently as Sunny Cove Intel added an instruction to speed up memory copying. AVX-512 is also shaping up to become very flexible compared to previous iterations, hopefully AMD will support it soon. Further down the line, Intel is researching into "threadlets", which have to potential for massive performance gains, but yet again are more in the "CISC" direction than "RISC".

Low hanging fruit and a clean slate. That is why ARM is so promising.

While ARM have changed a lot over the years, it dates back to ~1985, so I wouldn't call it quite a "clean slate".

 

[londiste]

ARM's promise is in lack of features and solutions that high-performance architectures have implemented over last few decades. ARM has been focused on low power and simply never needed anything of the sort. Today it is relatively easy for ARM to come within striking distance of other architectures.

 

[ARF]

Here is the history of Intel's progress with x86:

4-core Core i7 2600K 3.4 - 3.8 GHz (Sandy Bridge)
4-core Core i7 3770K 3.5 - 3.9 GHz (Ivy Bridge) + small IPC improvement
4-core Core i7 4770K 3.5 - 3.9 GHz (Haswell) + small IPC improvement
4-core Core i7 5775C 3.3 - 3.7 GHz (Broadwell) + small IPC improvement
4-core Core i7 6700K 4.0 - 4.2 GHz (Skylake) + small IPC improvement
4-core Core i7 7700K 4.2 - 4.5 GHz (Kaby Lake) + small IPC improvement

6-core Core i7 8700K 3.7 - 4.7 GHz (Coffee Lake) + small IPC improvement
8-core Core i9 9900K 3.6 - 5.0 GHz (Coffee Lake)

10-core Core i9 10900K 3.7 - 5.3 GHz (Comet Lake)

 

[notb]

Here is the history of Intel's progress with x86:

Your point being...? Is there any?
You don't like Intel. We get this.

 

[Vayra86]

Here is the history of Intel's progress with x86:

Before AMD's Ryzen

4-core Core i7 2600K 3.4 - 3.8 GHz (Sandy Bridge)
4-core Core i7 3770K 3.5 - 3.9 GHz (Ivy Bridge) + small IPC improvement
4-core Core i7 4770K 3.5 - 3.9 GHz (Haswell) + small IPC improvement
4-core Core i7 5775C 3.3 - 3.7 GHz (Broadwell) + small IPC improvement
4-core Core i7 6700K 4.0 - 4.2 GHz (Skylake) + small IPC improvement
4-core Core i7 7700K 4.2 - 4.5 GHz (Kaby Lake) + small IPC improvement

After AMD's Ryzen in 2017

6-core Core i7 8700K 3.7 - 4.7 GHz (Coffee Lake) + small IPC improvement
8-core Core i9 9900K 3.6 - 5.0 GHz (Coffee Lake)

10-core Core i9 10900K 3.7 - 5.3 GHz (Comet Lake)

Before AMD Ryzen
- My birth. 1986.

After AMD Ryzen
- My daughter's birth, 2018.

Thanks, AMD

 

[notb]

Thanks, AMD

Before AMD Ryzen:
I graduated, bought a flat, met my fiance, backpacked through big chunk of Europe.

After AMD Ryzen:
COVID-19

 

[ARF]

Thanks, AMD

I will edit my comment because you wrongly emphasize and focus on AMD when you didn't read the post actually.

10-core Comet Lake is basically Sandy Bridge with 2.5 times the cores, 1.4 times the IPC, some added complex instructions, and a mild frequency bump.
All in 9 years.

 

[Vayra86]

I will edit my comment because you wrongly emphasize and focus on AMD when you didn't read the post actually.

10-core Comet Lake is basically Sandy Bridge with 2.5 times the cores, 1.4 times the IPC, some added complex instructions, and a mild frequency bump.
All in 9 years.

Well yay, you want a cookie now? Does Intel need a cookie? x86?

Your point still eludes me. Is this somehow a good thing? Maybe you oughta look at core counts in, say, PowerPC over the years. The one you heralded with Cell... when was that, an 8 core in... 2005? And let's not begin about ARM... they had 64 core monstrosities years back.

 

[ARF]

Well yay, you want a cookie now? Does Intel need a cookie? x86?

Your point still eludes me. Is this somehow a good thing? Maybe you oughta look at core counts in, say, PowerPC over the years. The one you heralded with Cell... when was that, an 8 core in... 2005?

The headline of the article is x86 Lacks Innovation and you are arguing that x86 is a second coming of Jesus Christ.
x86 is something not optimal for us and we must move on to something better, otherwise more crisises will come.

 

[Vayra86]

The headline of the article is x86 Lacks Innovation and you are arguing that x86 is a second coming of Jesus Christ.
x86 is something not optimal for us and we must move on to something better, otherwise more crisises will come.

x86 is pretty optimal for me. It provides me with a job. An income. It allows me to play state of the art games and use top of the line hardware, that runs on one of the best OS'es in the world, supported by the most diverse and largest community on the planet.

ARM? Pretty lacking in that respect. It will need quite a few years to get there.

This goes back to my first comment in this topic, really. Its not ONLY about performance, its first and foremost about software. And software is the great equalizer here, too. Do you really think software will adjust to the limitations of ARM? Of course not. This brings up a dilemma: do we change ARM to make it usable with new demands in software? Or do we change the software to 'fit in' with ARM's limitations?

And just like that.... you've created another x86.

 

[notb]

x86 is something not optimal for us and we must move on to something better, otherwise more crisises will come.

So you agree Ryzen caused COVID-19 or what?

 

[bug]

For this in the know, Andrew Tanenbaum explained over 2 decades ago why x86 has become an inefficient hybrid in his Structured Computer Organization book.
Obviously, since then x86 was able to push boundaries with Core and Zen. But it also lost big time when trying to fit into the mobile space. The best solution is an ever shifting target, but x86's age is undeniably showing. It does not mean it's a bad architecture or anything (nothing that serves the public well for almost half a century ever is).

 

[londiste]

For this in the know, Andrew Tanenbaum explained over 2 decades ago why x86 has become an inefficient hybrid in his Structured Computer Organization book.

5.8.1 The Problem with the Pentium 4?

 

[ValenOne]

Modern X86 CPUs has RISC cores.

Ars Technica: RISC vs. CISC: the Post-RISC Era - Page 6 - (10/1999)

Ars Technica. Power users and the tools they love, without computing religion. Oh yeah, did we mention we are unassailable computing enthusiasts.

archive.arstechnica.com

Both the Athlon and the P6 run the CISC x86 ISA in what amounts to hardware emulation, but they translate the x86 instructions into smaller, RISC-like operations that fed into a fully post-RISC core. Their cores have a number of RISC features (LOAD/STORE memory access, pipelined execution, reduced instructions, expanded register count via register renaming), to which are added all of the post-RISC features we've discussed. The Athlon muddies the waters even further in that it uses both direct execution and a microcode engine for instruction decoding. A crucial difference between the Athlon (and P6) and the G4 is that, as already noted, the Athlon must translate x86 instructions into smaller RISC ops

 

[R0H1T]

Intel wants to grow and the lucrative mobile, IoT and networking markets are the most obvious directions.

Except mobile, IoT & networking, a lot of it, aren't really that lucrative. And the only ones who've consistently made a profit in the mobile sector are integrated players like Apple, Samsung & recently Huawei (for a lot of other reasons as well) so basically not Intel & they'll never be a force in this sector ever again.

 

[ValenOne]

You had to explain about IBM's PowerPC and the Sony's PlayStation 3 Cell.

Cell is a 2005-2006 technology and yet its raw performance is around Intel's current processors.

That's not true.

AMD GCN's has full support for IEEE-754-2008 similar SSE. AMD APU beats IBM CELL!

There's a reason why Sony has abandoned CELL!

AMD's PS5 APU (CPU has 896 GFLOPS full IEEE-754 support, GpGPU has 10.28 TFLOPS compute full IEEE-754-2008 support and 11 ~TFLOPS equivalent RT cores) will murder CELL.

Larger monster is XSX's APU (CPU has 972 GFLOPS full IEEE-754 support, GpGPU has 12.147 TFLOPS compute full IEEE-754-2008 support and 13 ~TFLOPS equivalent RT cores)

XSX's APU supports 97.176 TOPS (trillion operations per second) for machine learning AI.

Fusion beats CELL

 

[bug]

5.8.1 The Problem with the Pentium 4?

I obviously don't remember the exact chapter, but the gist of it was x86 is stuck between being a CISC architecture while trying to emulate a RISC one. His opinion was the hardware dedicated to this was unsustainable in the long term.

Edit: You made me look: in my 4th edition that's "5.8.1 The Problem with the Pentium II"

 

[ValenOne]

I obviously don't remember the exact chapter, but the gist of it was x86 is stuck between being a CISC architecture while trying to emulate a RISC one. His opinion was the hardware dedicated to this was unsustainable in the long term.

There are many false prophets against the X86.

AMD has K12 which is an ARMv8-A (AArch64) clone CPU.

AMD details Zen, K12 CPU cores, confirms next-gen memory support

AMD has confirmed details on its upcoming Zen and K12 processors, with a few additional tidbits on Fiji (its future GPU architecture) tossed in for good measure.

www.extremetech.com

AMD K12 R&D leverages much of the work done on x86.

 

[notb]

Except mobile, IoT & networking, a lot of it, aren't really that lucrative. And the only ones who've consistently made a profit in the mobile sector are integrated players like Apple, Samsung & recently Huawei (for a lot of other reasons as well) so basically not Intel & they'll never be a force in this sector ever again.

Again: you're talking about consumer products. Intel is into enterprise/industry solutions. Look up their Edge and 5G dedication.

Moreover, even if not as profitable as CPUs, it's still more profitable and closer to Intel's comfort zone than, let's say, producing games or selling insurance.
Intel is expected to grow. That's it. They're trying new things.

 

[ValenOne]

Guerrilla dev: PS3's Cell CPU is by far stronger than new Intel CPUs

Guerrilla dev: PS3's Cell CPU is by far stronger than new Intel CPUs

Sony's Cell architecture was beastly in power, and one dev says it's actually more powerful than today's Intel chips.

www.tweaktown.com

My PCs emulate PS3 just fine including Guerrilla's PS3 games.

 

[bug]

There are many false prophets against the X86.

AMD has K12 which is an ARMv8-A (AArch64) clone CPU.

AMD details Zen, K12 CPU cores, confirms next-gen memory support

AMD has confirmed details on its upcoming Zen and K12 processors, with a few additional tidbits on Fiji (its future GPU architecture) tossed in for good measure.

www.extremetech.com

AMD K12 R&D leverages much of the work done on x86.

I don't think anyone is against x86. But when you're doing CPU design, you'd better get your bearings straight, that's all.

 

[ValenOne]

Here is the history of Intel's progress with x86:

4-core Core i7 2600K 3.4 - 3.8 GHz (Sandy Bridge)
4-core Core i7 3770K 3.5 - 3.9 GHz (Ivy Bridge) + small IPC improvement
4-core Core i7 4770K 3.5 - 3.9 GHz (Haswell) + small IPC improvement
4-core Core i7 5775C 3.3 - 3.7 GHz (Broadwell) + small IPC improvement
4-core Core i7 6700K 4.0 - 4.2 GHz (Skylake) + small IPC improvement
4-core Core i7 7700K 4.2 - 4.5 GHz (Kaby Lake) + small IPC improvement

6-core Core i7 8700K 3.7 - 4.7 GHz (Coffee Lake) + small IPC improvement
8-core Core i9 9900K 3.6 - 5.0 GHz (Coffee Lake)

10-core Core i9 10900K 3.7 - 5.3 GHz (Comet Lake)

Note that "X86-64" instruction set holder is AMD not Intel LOL.

Intel is the cloner for AMD's X86-64 CPUs.

 

[R0H1T]

Again: you're talking about consumer products. Intel is into enterprise/industry solutions. Look up their Edge and 5G dedication.

When you're talking mobile ~ Intel is dead there, worse than a dodo be it consumer or enterprise sector!

IoT ~ evolving space & a security nightmare, remind us what's their achievement there?

Also for networking how about not patching routers(?) with massive security holes

Intel Coughs to Puma CPU Flaw that Hit Virgin Media Hub 3 Router UPDATE

After nearly two years Intel has finally published an advisory and formal CVE entry for a flaw in their Puma 5, 6 and 7 chipsets that resulted in various broadb

www.ispreview.co.uk

Networking ~ that's the only space where there are competitive & in some ways leading, part because Broadcom & QC are both high end & command price premiums!

 

[qcmadness]

Again: you're talking about consumer products. Intel is into enterprise/industry solutions. Look up their Edge and 5G dedication.

Moreover, even if not as profitable as CPUs, it's still more profitable and closer to Intel's comfort zone than, let's say, producing games or selling insurance.
Intel is expected to grow. That's it. They're trying new things.

Intel is so strong that it sold its baseband solution to Apple.

For IoT, car and other appliance solutions, a quad-core generic ARM cores will do.
1. If you go x86, you will risk to have only 2 suppliers (Intel / AMD) while when you go ARM, you can just license and create your own solution.
2. x86 has neither performance nor power efficiency benefits over ARM in this segment.

There are many examples that Intel does not want to go "low-end":
1. Optane over NAND flash (sold to Micron)
2. x86 tablet solutions (ARM is dominant)
3. Rambus over DDR-RAM

 

[efikkan]

There are many false prophets against the X86.

Yeah, people have been predicting the end of x86 since the early 80s.
It's even mentioned in movies; RISC architecture is gonna change everything
But on a serious note, nothing is going to challenge x86 performance wise in the near future, and when something eventually does, it will not be ARM, it will be something more SIMD oriented, or a brand new concept none of us can concieve yet.

AMD has K12 which is an ARMv8-A (AArch64) clone CPU.

AMD details Zen, K12 CPU cores, confirms next-gen memory support

AMD has confirmed details on its upcoming Zen and K12 processors, with a few additional tidbits on Fiji (its future GPU architecture) tossed in for good measure.

www.extremetech.com

AMD K12 R&D leverages much of the work done on x86.

Most have forgotten that K12 was supposed to be the next big archtecture after Bulldozer, Zen was intended as a stop-gap, but since then K12 has "disappeared" and AMD plans to keep iterating Zen for many years to come. If K12 is not completely abandoned, it probably exists as a low-priority research project. Back in those days AMD had big plans to abandon x86, even with their failed "project skybridge" to have platforms supporting both x86 and ARM.