Imagination introduces Catapult RISC-V CPU cores
As expected, Imagination Technologies is giving another try to the CPU IP market with the Catapult RISC-V CPU cores following their previous unsuccessful attempt with the MIPS architecture, notably the Aptiv family.
Catapult RISC-V CPUs are/will be available in four distinct families for dynamic microcontrollers, real-time embedded CPUs, high-performance application CPUs, and functionally safe automotive CPUs.
The new 32-/64-bit RISC-V cores will be scalable to up to eight asymmetric coherent cores-per cluster, offer a "plethora of customer configurable options", and support optional custom accelerators. What you won't see today are block diagrams and detailed technical information about the cores because apparently, all that information is confidential even though some Catapult RISC-V cores are already shipping "in high-performance Imagination automotive GPUs". The only way to get more details today is to sign an NDA.
Also at AnandTech and Phoronix.
Previously: Imagination Announces B-Series GPU IP: Scaling up with Multi-GPU
Imagination Technologies Plans to Design RISC-V Cores
Related: Innosilicon Graphics Cards Based on "Fantasy One" GPU Feature Up to 32GB GDDR6X Memory
Related Stories
Imagination has announced new B-series GPU designs focused on automotive and high-performance computing use cases, as it has become difficult for the company to compete in the mobile GPU market:
It's almost been a year since Imagination had announced its brand-new A-series GPU IP, a release which at the time the company called its most important in 15 years. The new architecture indeed marked some significant updates to the company's GPU IP, promising major uplifts in performance and promises of great competitiveness. Since then, other than a slew of internal scandals, we've heard very little from the company – until today's announcement of the new next-generation of IP: the B-Series.
The new Imagination B-Series is an evolution of last year's A-Series GPU IP release, further iterating through microarchitectural improvements, but most importantly, scaling the architecture up to higher performance levels through a brand-new multi-GPU system, as well as the introduction of a new functional safety class of IP in the form of the BXS series.
[....] Imagination's current highest-end hardware implementation in the BXT series is the BXT 32-1024, and putting four of these together creates an MC4 GPU. In a high-performance implementation reaching up to 1.5GHz clock speeds, this configuration would offer up to 6TFLOPs of FP32 computing power. Whilst this isn't quite enough to catch up to Nvidia and AMD, it's a major leap for a third-party GPU IP provider that's been mostly active in the mobile space for the last 15 years.
[....] Beyond the addition of safety critical features on the BXS series, the automotive IP also features some specific enhancements in the microarchitecture that allows for better performance scaling for workloads that are more unique to the automotive space. One such aspect is geometry, where automotive vendors have the tendency to use absurd amounts of triangles. Imagination says they've tweaked their designs to cover these more demanding use-cases, and together with some MSAA specific optimisations they can reach up to a 60% greater performance for these automotive edge-cases, compared to the regular non-automotive IP.
Related: Imagination Technologies Group Up for Sale
Imagination Technologies to design RISC-V cores:
Now better known for its PowerVR embedded GPUs, Imagination Technologies tried to enter the CPU market by purchasing MIPS Technologies and introducing microAptiv, interAptiv, and proAptiv cores in 2012.
It did not end up well, as the company had to sell its MIPS technology a few years later, and the MIPS architecture is now barely supported. But Imagination is now working on getting back into the CPU space by designing RISC-V cores.
[...] a May 2021 report by the Financial Times claims Imagination expects to invest up to $150m over the next two years to target a fresh push into the processor design market, specifically citing the RISC-V architecture.
Also at Tom's Hardware.
See also: QEMU 6.1 Released With RISC-V Improvements, AMD Emulation Fixes
Related: Imagination Technologies Acquired for $675 Million, MIPS to be Sold Off
Wave Computing Acquires MIPS Technologies
Imagination Announces B-Series GPU IP: Scaling up with Multi-GPU
Innosilicon graphics cards based on "Fantasy One" GPU feature up to 32GB GDDR6X memory
Today at the "Fantasy One GPU Product Press Conference", Innosilicon, a Chinese company offering graphics and memory solutions unveiled its first discrete GPU.
At the event, Innosilicon revealed its plans for the Fantasy One GPU. This processor is based on Imagination graphics IP (BXT to be specific) which brings tons of innovations to the discrete GPU solutions offered by the Chinese company.
As many as four products have been revealed, including dual-GPU and single-GPU solutions. Type A is a consumer/workstation graphics card featuring a single Fantasy One GPU, which is actually a multi-chip (chiplet) design. Unfortunately, none of the news reports from China on this announcement had the exact configuration of the Fantasy One.
According to the data provided by Innoslicon, this GPU offers up to 5 TFLOPS of single-precision compute power [and a fillrate of 160 GPixel/s]. It is equipped with DisplayPort 1.4, HDMI 2.1, and VGA connectors. This card is to be equipped with up to 16 GB of GDDR6(X) memory across a 128-bit interface. So far, the G6X technology has been exclusive to NVIDIA/Micron Ampere GPUs, but apparently, Innosilicon made a lot of research in PAM4 signaling and was able to unlock up to 19 Gbps memory bandwidth for their GDDR6X implementation. A relatively short memory bus will have its toll on the maximum theoretical bandwidth though, which is to go up to 304 GB/s (so somewhere in between Radeon RX 6700XT and 6600XT).
[...] The Type B [...] is a dual-GPU solution featuring two Fantasy One GPUs connected by an interface known as Innolink. The company claims up to 10 TFLOPS of computing power and 320 GPixel/s fillrates. This card can offer 32 simultaneous 1080p/60FPS streams or 64 streams at 720/30FPS. It is to feature up to 32GB of GDDR6(X) memory but is again limited by dual 128-bit interfaces from each GPU. All these cards feature a PCI-Express 4.0 interface at full X16 width.
When people are willing to pay 2-4x as much for GPUs, competition is inevitable. Intel will be joining the market with "Alchemist" discrete GPUs in Q1/Q2 2022.
Previously: Imagination Announces B-Series GPU IP: Scaling up with Multi-GPU
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @02:39PM (3 children)
the confidentiality puts the whole thing at risk.
(Score: 2) by Runaway1956 on Tuesday December 07 2021, @02:54PM (2 children)
a Beowulf cluster of these clusters!
Or, is that old joke worn out now?
Abortion is the number one killed of children in the United States.
(Score: 3, Touché) by DannyB on Tuesday December 07 2021, @04:37PM (1 child)
I for one welcome our new RISC-V overlords.
If you think a fertilized egg is a child but an immigrant child is not, please don't pretend your concerns are religious
(Score: 1, Insightful) by Anonymous Coward on Tuesday December 07 2021, @05:48PM
with lasers, sharks, and hot grits (or great tits - a salute to TMB)
(Score: 1, Interesting) by Anonymous Coward on Tuesday December 07 2021, @03:03PM (8 children)
i don't want to pooh-pooh on the parade ... but how much "chip" is needed for a electric car you don't mind driving yourself? also how many "bomble"-flops do you really need to carry around to make a phone call?
it seems all the extra transistors go into advertising your need for mEOr " bomble"-flops.
today i found (and bought) a solid state relay. it has a dc input 9v-25v that can switch 135-250VAC/50hz @ 40A current on and off.
so basically this match-box size device with "half a nano bomble flop" now controls if my hybrid inverter is connected to the grid or not.
@ dc input there's a mini solarpanel that controls the relay. if it "sees" sun, the ac side closes and connects the hybrid inverter to grid. if it's dark the hybrid inverter and battery turn into off-grid. thanks to half-a-nano bomble flop of processing... the grid is now my "excess energy" dumping ground. like a landfill, junk yard but for electricity (we don't want to overcharge/blow up) the battery, lol.
(Score: 2, Interesting) by Anonymous Coward on Tuesday December 07 2021, @04:20PM (1 child)
> how many "bomble"-flops do you really need to carry around to make a phone call?
I was trying to reduce active (not idle) power consumption* on my phone (playing podcasts burns through the battery at around 3% per hour), and experimented with disabling cpu cores.
It has a Snapdragon 730G A55 and A76 cores.
I disabled all but one A55 "small" core, and the phone was very painful to use. But, with two A55 cores, web browsing (with ublock origin), calls, messaging, UI navigation, podcast player, etc., were all fine. So, for my use case, I apparently don't need anywhere close to the maximum "bomble-flops" available on my phone.
*Even with only a single A55 core enabled, it didn't help with active power consumption, though.
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @04:34PM
Phone manufacturers have tuned the hell out of basic power efficiency at this point for RUNNING software.
The only thing you can do to significantly increase your power efficiency is turn down your screen brightness and disable software/services that you don't give a damn about but that the phone manufacturer or software companies do, generally for the purposes of spying on you to collect marketing data.
(Score: 0, Disagree) by Anonymous Coward on Tuesday December 07 2021, @04:32PM (2 children)
This doesn't make you sound cool, intelligent, aloof, or whatever you were going for.
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @06:17PM
sorry ...
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @09:03PM
To better help our British editors understand, I propose we henceforth refer to them as "boffin-flops".
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @04:38PM
Physical control loops generally only need to operate at a frequency of no more than 1 kHz. Imagine a processor dedicated to operating that control loop. A CPU frequency of 1 to 10 MHz might be enough to run that 1 kHz physical update control loop.
(Score: 4, Informative) by DannyB on Tuesday December 07 2021, @04:43PM
<no-sarcasm>
Big systems with big applications want lots of cpu power and lots of memory. Especially if written in Java.
</no-sarcasm>
It took Deep Thought 7-1/2 million years (according to this [wikipedia.org]) to compute the ultimate answer to life the universe and everything.
Imagine how much time could have been saved by a 20% increase in cpu performance?
If you think a fertilized egg is a child but an immigrant child is not, please don't pretend your concerns are religious
(Score: 2) by takyon on Tuesday December 07 2021, @10:06PM
https://www.cnx-software.com/2021/12/06/amlogic-v901d-a-quad-core-cortex-a55-processor-for-in-vehicle-infotainment/ [cnx-software.com]
The new minimum for infotainment!
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 1, Interesting) by Anonymous Coward on Tuesday December 07 2021, @06:49PM (1 child)
Imagination, as per company policy it seems, is difficult to work with. They sure don't want to supply documentation and they sure don't want you to talk about their products if you're unfortunate enough to have purchased any.
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @08:47PM
Company has been circling the drain for years since Apple dumped them.
Today, PowerVR is synonymous for cheap Android phones that Mediatek got a sweetheart deal not to include Mali on.
FOSS drivers exist for most other GPUs except this one - FSF still has a high priority page to reverse engineer the firmware from back when OMAP was still relevant more than a decade ago.
Avoid.
(Score: 0) by Anonymous Coward on Tuesday December 07 2021, @09:56PM
"What you won't see today are block diagrams and detailed technical information about the cores because apparently, all that information is confidential even though some Catapult RISC-V cores are already shipping "in high-performance Imagination automotive GPUs". The only way to get more details today is to sign an NDA."
Imagination are stupid slaveware peddling whores.
(Score: 0) by Anonymous Coward on Wednesday December 08 2021, @12:30AM
Imagination Technologies Announces "Catapult" RISC-V CPU Cores
If it’s made by Imagination, it might as well be “Catapulted” right into the dumpster.