Credit: AheadComputing
Intel, CPUs, and the concept of "badness" aren't necessarily things you'd want to shout about, what with numerous well-documented issues afflicting Intel's recent processors. But a new Oregon-based startup called AheadComputing is leaning hard on the Intel provenance of its founders while claiming that it is creating, "the biggest, baddest CPU in the world." And it's going to do it via IPC or instructions per clock, not cranking up the operating frequencies or throwing in more cores.
That is some statement. All four of AheadComputing's founders had long careers at Intel, dating all the way back to ye olde 386 processor through to the latest Intel Core-branded chips. What's more, AheadComputing also appointed CPU design legend Jim Keller to its board in March. That's at least a vote of confidence, even if it seems unlikely Keller will be involved in the design of AheadComputing's CPUs.
The company is very young, having launched in July last year with a plan to, "develop and license breakthrough, high-performance 64-bit RISC-V processor cores." RISC-V, of course, is an open-source instruction set that exists to present a more modern and cost effective alternative to the proprietary x86 and Arm standards.
Currently, RISC-V chips tend to be found in embedded applications and commercial devices. RISC-V has yet to make much of an impact in PCs or phones, for instance.
Exactly how AheadComputing is going to deliver on that promise of the "biggest, baddest CPU in the world" isn't totally clear beyond the focus in IPC. It's a fabless startup, which means it won't manufacture chips itself. But then the likes of AMD and Nvidia are fabless, too. It's really only Intel that designs and manufactures its own chips, and that business is coming under increasing pressure.
According to AheadComputing's CEO Debbie Marr, "the x86 ecosystem is fiercely defending its territory but is destined to lose in the end." As for Arm, she says, "we anticipate that the ARM ecosystem will experience considerable strain in the coming years. If ARM's current customers are pressured excessively, they will consider transitioning to an alternative architecture like RISC-V."
In response, AheadComputing claims it will, "demonstrate leadership in CPU performance and performance per watt in a very short timeframe and start building the second generation of products that will demonstrate our commitment to a roadmap with large gains in performance generation over generation."
AheadComputing says it will achieve that via IPC, or instructions processed per clock, as opposed to operating frequency or adding cores. "If the performance and efficiency from the multi-core scaling era are slowing down, then it's time for the CPU designers to find a different way to use the additional gates from new process technologies. CPU designers must look towards IPC. This will require increasing the functions for each core rather than increasing the number of cores. If we do this intelligently, AheadComputing will provide performance improvements regardless of workload parallelism," says co-founder Jonathan Pearce.
That latter point could be critical. When Intel's plans for 10 GHz-plus computing hit the wall towards the latter end of the 2000's, the company dramatically changed tack in favour of multi-core computing as a way to add performance in the absence of substantial clockspeed improvements.
Adding lots of cores isn't always the best way to improve performance. | Credit: Intel
The problem with adding cores is that it relies on multi-threaded workloads. That's fine for many tasks, like 3D rendering. But it's not a magic bullet for every computational task. Indeed, that's why AMD's eight-core Ryzen 7 9800X3D is the weapon of choice for PC gaming, currently. Adding another eight cores in the form of the Ryzen 9 9950X3D typically doesn't do a whole lot for gaming performance.
Whatever, aside from that focus on IPC as opposed to adding cores, AheadComputing isn't going into any detail. For sure, it will be years before the company's CPU core designs have any chance of showing up in a device you can actually buy.
Personally, if you offered me a CPU with either 50% more IPC or 50% more cores, I'd take the IPC every time. That will deliver in almost any circumstance, while multi-core CPUs can be a bit more hit and miss. Aiming for improved IPC also tends to make for better efficiency, which is great for mobile PCs.
Anywho, for now we'll have to chalk AheadComputing down as a slow burn. The company has strong provenance, but it's anyone's guess as to whether it will, in reality, make an impact. My best guess is that if it manages to come up with an interesting core design, it'll get snapped up by one of the big boys, just as the startup Nuvia was bought by Qualcomm and its Oryon CPU cores ended up in the new Snapdragon X chips.
And all of that is before you even begin to ponder the odds of any RiSC-V chip making an impact on the PC. Industry watchers have been predicting Arm chips would take over the PC for decades. That still hasn't happened.
Ex-Intel engineers are developing the 'biggest, baddest CPU in the world' by targeting IPC, not clockspeed or core counts
AheadComputing reckons the future of CPU performance is all about instructions per clock, not GHz or a zillion cores.
