Discussion RISC V Latest Developments Discussion [No Politics]

Page 13 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DisEnchantment

Golden Member
Mar 3, 2017
1,774
6,757
136
Some background on my experience with RISC V...
Five years ago, we were developing a CI/CD pipeline for arm64 SoC in some cloud and we add tests to execute the binaries in there as well.
We actually used some real HW instances using an ARM server chip of that era, unfortunately the vendor quickly dumped us, exited the market and leaving us with some amount of frustration.
We shifted work to Qemu which turns out to be as good as the actual chips themselves, but the emulation is buggy and slow and in the end we end up with qemu-user-static docker images which work quite well for us. We were running arm64 ubuntu cloud images of the time before moving on to docker multi arch qemu images.

Lately, we were approached by many vendors now with upcoming RISC-V chips and out of curiosity I revisited the topic above.
To my pleasant surprise, running RISC-V Qemu is smooth as butter. Emulation is fast, and images from Debian, Ubuntu, Fedora are available out of the box.
I was running ubuntu cloud images problem free. Granted it was headless but I guess with the likes of Imagination Tech offering up their IP for integration, it is only a matter of time.

What is even more interesting is that Yocto/Open Embedded already have a meta layer for RISC-V and apparently T Head already got the kernel packages and manifest for Android 10 working with RISC-V.
Very very impressive for a CPU in such a short span of time. What's more, I see active LLVM, GCC and Kernel development happening.

From latest conferences I saw this slide, I can't help but think that it looks like they are eating somebody's lunch starting from MCUs and moving to Application Processors.


And based on many developments around the world, this trend seems to be accelerating greatly.
Many high profile national and multi national (e.g. EU's EPI ) projects with RISC V are popping up left and right.
Intel is now a premium member of the consortium, with the likes of Google, Alibaba, Huawei etc..
NVDA and soon AMD seems to be doing RISC-V in their GPUs. Xilinx, Infineon, Siemens, Microchip, ST, AD, Renesas etc., already having products in the pipe or already launched.
It will be a matter of time before all these companies start replacing their proprietary Arch with something from RISC V. Tools support, compiler, debugger, OS etc., are taken care by the community.
Interesting as well is that there are lots of performant implementation of RISC V in github as well, XuanTie C910 from T Head/Alibaba, SWerV from WD, and many more.
Embedded Industry already replaced a ton of traditional MCUs with RISC V ones. AI tailored CPUs from Tenstorrent's Jim Keller also seems to be in the spotlight.

Most importantly a bunch of specs got ratified end of last year, mainly accelerated by developments around the world. Interesting times.
 

Doug S

Diamond Member
Feb 8, 2020
3,120
5,361
136
The reported SPECint scores for Ascalon don't really match up with "Projected Zen5 performance in 2024".
Callandor looks insane though, 16 wide decode with 2-ahead branch predictor and SPECint2017/GHz 40% above the M4.

That sounds impressive and I'm not saying they can't do it, but hyper focusing on IPC and not mentioning actual performance anywhere seems like they're hiding something.

It is always easier to get higher IPC at lower clock rates, so if they designed something that has a ceiling of say 3 GHz they might get incredible IPC but still fall well short in terms of performance. A lot fewer people were impressed with Apple's iPhone SoCs lofty IPC when they could barely muster half the actual performance of Intel's best.

So where's the graph that shows where they will be in 2027 versus where Apple, Intel and AMD are today? Back when Nuvia was making lofty claims they didn't show us a graph of IPC, they showed us a graph of Geekbench scores and where they were shooting for.
 
Reactions: Nothingness

soresu

Diamond Member
Dec 19, 2014
3,708
3,037
136
Full slides here: https://riscv.or.jp/wp-content/uploads/Japan_RISC-V_day_Spring_2025_compressed.pdf

View attachment 118540View attachment 118538
View attachment 118537View attachment 118536

The reported SPECint scores for Ascalon don't really match up with "Projected Zen5 performance in 2024".
Callandor looks insane though, 16 wide decode with 2-ahead branch predictor and SPECint2017/GHz 40% above the M4.
RVA25 in 2027 seems a bit presumptuous considering RVA23 just went gold a couple of months ago.
 

soresu

Diamond Member
Dec 19, 2014
3,708
3,037
136
That sounds impressive and I'm not saying they can't do it, but hyper focusing on IPC and not mentioning actual performance anywhere seems like they're hiding something.
This far out (2+ years) they can only be using simulation - not an actual engineering sample.

So they have literally nothing to hide.
 
Jul 27, 2020
24,129
16,830
146
I'm sure he and the rest of the crew at Tenstorrent will take talent regardless of its origin.
Maybe. But based on what adroc has said in the past, Keller's ideas weren't welcomed by IDC and my own assumption is because they are arrogant. But oh well. Good if Keller's current team made up of whoever manages to upstage Apple's crack team of designers.
 

jdubs03

Golden Member
Oct 1, 2013
1,232
871
136
Maybe. But based on what adroc has said in the past, Keller's ideas weren't welcomed by IDC and my own assumption is because they are arrogant. But oh well. Good if Keller's current team made up of whoever manages to upstage Apple's crack team of designers.
I’ll just say: Consider me highly skeptical. Even Nuvia needed former Apple experts to get close. But I stand to be corrected if so.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,803
1,286
136
RVA25 in 2027 seems a bit presumptuous considering RVA23 just went gold a couple of months ago.
RVA25 isn't going to be the final thingamabob.

RVA23 = RVA23.0
RVA24 = RVA23.1
RVA25 = RVA23.2
It will eventually switch to RVA30.x, then to RVA40.x since they couldn't freeze yearly.
 

Nothingness

Diamond Member
Jul 3, 2013
3,292
2,357
136
Israelis are arrogant. I'm sure Keller would prefer not to deal with that, unless he has no other choice.
This kind of generalisation is quite ironic coming from an American at this moment in time. Even if Intel Israel engineers might have been arrogant, that can't be translated to Israelis are arrogant.

Callandor looks insane though, 16 wide decode with 2-ahead branch predictor and SPECint2017/GHz 40% above the M4.
How do you come to 40%? Tenstorrent said 3.5 SPEC17/GHz. M4 was publicly measured at >3 SPEC17/GHz. That'd mean +15%.

I'm a bit surprised by the 16-wide decoder along with uop cache. R-V needs a significant amount of instruction fusion to reduce uop bandwidth, but I would have expected the uop cache to store fused uops, thus reducing the need for very wide decode. Looking at the diagram, I wonder if it's not only to take care of compressed code (32-byte fetch + aligner seem to hint at that).
 

Thibsie

Golden Member
Apr 25, 2017
1,031
1,206
136
This kind of generalisation is quite ironic coming from an American at this moment in time. Even if Intel Israel engineers might have been arrogant, that can't be translated to Israelis are arrogant.

It's not only ironic, it is dangerous and far from showing any smartness.
 

naukkis

Golden Member
Jun 5, 2002
1,004
844
136
So Callandor will be clustered design with shared register file. Something very Keller like design from past, which I might have predicted will be coming. Being clustered design does mean that it will clock higher than similar non-clustered design - if build well it should clock just as 8-wide designs.

It's quite interesting that both Intel and AMD did use Kellers visions just to build better x86 cpu instead of targeting best performance possible. Intel even did try to do clustered x86.... So in few years they will face at least two Keller's clustered design rv rivals, one from Tenstorrent and one from Intels Royal Cove team, which might have found that instead their design being bad it might have been just bad ISA instead. I think that Arm inc should also hire Keller to make new directions for their ISA - they really seem to lost focus lately. ( Keller probably influenced original Aarch64 decisions.....)
 
Last edited:

Nothingness

Diamond Member
Jul 3, 2013
3,292
2,357
136
Phoronix benchmarked speed improvements going from Ubuntu 21 to 24 on Hifive Unmatched board running an U740 CPU.

Summary (take with care, I only use the aggregated score, look at individual tests for a more to the point comparison):
1. Ubuntu 24 is 23% faster than Ubuntu 21; most of the speedup seems to come from maturing compilers
2. The R-V board performance is ~10% that of a Pi 500.

I insist, take these results with care.

 
Last edited:

camel-cdr

Member
Feb 23, 2024
27
90
51
This kind of generalisation is quite ironic coming from an American at this moment in time. Even if Intel Israel engineers might have been arrogant, that can't be translated to Israelis are arrogant.


How do you come to 40%? Tenstorrent said 3.5 SPEC17/GHz. M4 was publicly measured at >3 SPEC17/GHz. That'd mean +15%.

I'm a bit surprised by the 16-wide decoder along with uop cache. R-V needs a significant amount of instruction fusion to reduce uop bandwidth, but I would have expected the uop cache to store fused uops, thus reducing the need for very wide decode. Looking at the diagram, I wonder if it's not only to take care of compressed code (32-byte fetch + aligner seem to hint at that).
I used the numbers from the M4 Geekerwan video, which was 11.72@4.47GHz.
 

MS_AT

Senior member
Jul 15, 2024
599
1,253
96
So in few years they will face at least two Keller's clustered design rv rivals, one from Tenstorrent and one from Intels Royal Cove team, which might have found that instead their design being bad it might have been just bad ISA instead.
It is funny, if you have so much respect for Keller you will remeber he said that ISA wars are effectively waste of time and they don't matter

The only reason he turned to RISC-V was the licensing or lack of it to be more precise.

And in all fairness I think you are overstating contributions he makes to the designs themselves based on what he himself claims.
 

Nothingness

Diamond Member
Jul 3, 2013
3,292
2,357
136
And in all fairness I think you are overstating contributions he makes to the designs themselves based on what he himself claims.
It's incredibly naive to think a single man can make a CPU design outstanding. Anyone thinking that has obviously never worked on a successful design. It takes many brilliant people to achieve that.

I'm sure Keller has a solid and valid vision of what a CPU uarch should be; but I bet half of what he proposes is plain wrong and useless, because that's how things work: you propose high level ideas long in advance based on your experience, then the team hits the reality wall and starts iterating, sometimes dismissing these HL ideas, proposing new things and converging to something that works well.

Keller is bright for sure, but thinking everything he touches turns to gold is silly.
 

naukkis

Golden Member
Jun 5, 2002
1,004
844
136
And in all fairness I think you are overstating contributions he makes to the designs themselves based on what he himself claims.

I have questioned why nobody haven't do clustered execution cpu design. Last commercial one, Alpha 21264 Keller was part of design team and by this time its pretty clear that Keller has vision for clustered designs - Intel Royal and now that Callandor and those Ahead computing cpu. Those are only projected performance figures - but oh boy doesn't they look good? Keep in mind that Tenstorrent design is pretty much shoelaces budget against what Intel could spend on cpu design. Wonder about that AMD K12 design - there's not much details about it further than Kellers "it's got a bigger engine in it" - was it also meant to be two cluster version of zen execution hardware?
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |