Is there any reason why IBM doesn't sell 'consumer' CPUs?

[chrisjames61]

68040 is a 68000, not a PPC. Radically different things.

No kidding. I was replying to the guy who asked about the cpu in the NeXT machines.

 

[NTMBK]

Of course with PowerPC now being open source, perhaps someone new will come along with a consumer PowerPC CPU. I'd love to see modern Apple design chops plus an open PowerPC license.

 

[whm1974]

Of course with PowerPC now being open source, perhaps someone new will come along with a consumer PowerPC CPU. I'd love to see modern Apple design chops plus an open PowerPC license.

So we can now use the PowerPC without permission from IBM or payment of Royalties?

 

[SarahKerrigan]

So we can now use the PowerPC without permission from IBM or payment of Royalties?

Yes. That was announced a couple of weeks ago.

 

[whm1974]

Yes. That was announced a couple of weeks ago.

Great, but we can assume that IBM's designs are not Open?

 

[SarahKerrigan]

Great, but we can assume that IBM's designs are not Open?

Not at present. The only company to have released the RTL to a current flagship server processor that I'm aware of is Sun with the T1 and T2.

 

[whm1974]

Not at present. The only company to have released the RTL to a current flagship server processor that I'm aware of is Sun with the T1 and T2.

I wonder if it would be worthwhile to come up with new processor designs using SPARC and PPC ISAs? I've PowerPC CPUs have decent Performance Per Watt, does it?

 

[SarahKerrigan]

I wonder if it would be worthwhile to come up with new processor designs using SPARC and PPC ISAs? I've PowerPC CPUs have decent Performance Per Watt, does it?

Perf/W is largely agnostic to ISA unless the ISA is mis-designed. I'm not sure why anyone would do a new microarchitecture on SPARC, though, given that the SPARC ISA has not (imo) aged terribly well.

PPC has its good points, and its bad points. More good than bad, I think. I wish it had proper standardized embedded-type SIMD, (like Freescale SPE more than VMX) and a standard variable-length extension that people actually use, but I generally find it enjoyable to work with.

 

[whm1974]

Perf/W is largely agnostic to ISA unless the ISA is mis-designed. I'm not sure why anyone would do a new microarchitecture on SPARC, though, given that the SPARC ISA has not (imo) aged terribly well.

PPC has its good points, and its bad points. More good than bad, I think. I wish it had proper standardized embedded-type SIMD, (like Freescale SPE more than VMX) and a standard variable-length extension that people actually use, but I generally find it enjoyable to work with.

What are your thoughts about the RISC-V ISA? I know only a few SoCs are around such as one from SiFive but I been sort of watching it off and on for a while.

 

[SarahKerrigan]

What are your thoughts about the RISC-V ISA? I know only a few SoCs are around such as one from SiFive but I been sort of watching it off and on for a while.

As I've mentioned in previous threads, I am not overly impressed. It's not terrible but it leaves very useful stuff (integer rotate, clz, pre/post-increment) out of the base ISA - and, at present, outside of the standard extensions. (Yes, I know BitManip is coming.)

 

[whm1974]

As I've mentioned in previous threads, I am not overly impressed. It's not terrible but it leaves very useful stuff (integer rotate, clz, pre/post-increment) out of the base ISA - and, at present, outside of the standard extensions. (Yes, I know BitManip is coming.)

From I understand there a large number of companies that are members of the foundation that was setup for development of the ISA. As far as I know it is currently used mosty for training students.

 

[Eug]

Zombie thread! Brings back memories. I owned a iMac G5 with an IBM PowerPC 970FX. BTW, these were user serviceable iMacs with beautiful clean designs inside.

Apple says its iMac G5 is fully user-serviceable | AppleInsider

Apple\'s claims that its new iMac G5 computer is the worlds thinnest desktop computer, but it may also be the most user-serviceable.

appleinsider.com

Too bad they made the Intel iMacs totally unserviceable by end users.

Of course with PowerPC now being open source, perhaps someone new will come along with a consumer PowerPC CPU. I'd love to see modern Apple design chops plus an open PowerPC license.

Nah. They've thrown their money and might all behind ARM, and they already have desktop class cores.

Their latest purported A13 SoC gets a Geekbench score of 5415, with an integer score of 6110, while running at 2.66 GHz... in an iPhone.

Apple A13: Launch 2019-09-10 - Geekbench 4 & 5 scores available (GB4 multi-core 14070)

I just found this entry for iPhone12,1 on Geekbench. If real, that would make it the 2019 iPhone with A13: https://browser.geekbench.com/v4/cpu/14528580 Single-core 5415, which would be about 12% faster than the fastest A12 score posted. The multi-core score of 11294 isn't very impressive...

forums.anandtech.com

 

[NTMBK]

Zombie thread! Brings back memories. I owned a iMac G5 with an IBM PowerPC 970FX. BTW, these were user serviceable iMacs with beautiful clean designs inside.

Apple says its iMac G5 is fully user-serviceable | AppleInsider

Apple\'s claims that its new iMac G5 computer is the worlds thinnest desktop computer, but it may also be the most user-serviceable.

appleinsider.com

Too bad they made the Intel iMacs totally unserviceable by end users.


Nah. They've thrown their money and might all behind ARM, and they already have desktop class cores.

Their latest purported A13 SoC gets a Geekbench score of 5415, with an integer score of 6110, while running at 2.66 GHz... in an iPhone.

Apple A13: Launch 2019-09-10 - Geekbench 4 & 5 scores available (GB4 multi-core 14070)

I just found this entry for iPhone12,1 on Geekbench. If real, that would make it the 2019 iPhone with A13: https://browser.geekbench.com/v4/cpu/14528580 Single-core 5415, which would be about 12% faster than the fastest A12 score posted. The multi-core score of 11294 isn't very impressive...

forums.anandtech.com

Yeah, but they have to pay license fees to ARM. It's not like their design team doesn't have PowerPC experience- they initially started out as P.A. Semi, making the Pwrficient CPU.

I know it's a long shot, I just think it would be cool to see PowerPC come back

 

[whm1974]

Yeah, but they have to pay license fees to ARM. It's not like their design team doesn't have PowerPC experience- they initially started out as P.A. Semi, making the Pwrficient CPU.

I know it's a long shot, I just think it would be cool to see PowerPC come back

Yeah but wouldn't PowerPC have a lot of catching up to do? I mean aside from AmigaOS Boxes PPC hasn't in desktop systems for at least over a decade.

 

[moinmoin]

Apple is in the unique position of controlling the whole software and hardware stack, so they could eventually move away from the ARM ISA with most people not noticing. Apple already has an Architecture License, so they don't necessarily use ARM design modules or "the" ARM ISA but actually can already adapt and change it to their needs (including moving away from what makes ARM ARM). On the other hand Apple works together with ARM/Acorn since the Newton days, so the situation is not comparable to Imagination.

 

[SPBHM]

the last consumer product with a PowerPC was probably the Wii U (2012)? (and it used some variation of the gamecube CPU, stuff from the late 90s I think)

 

[NTMBK]

Yeah but wouldn't PowerPC have a lot of catching up to do? I mean aside from AmigaOS Boxes PPC hasn't in desktop systems for at least over a decade.

I presume Apple would start from their current ARM designs, as opposed to old PPC cores.

 

[whm1974]

I presume Apple would start from their current ARM designs, as opposed to old PPC cores.

I doubt Apple will go back to PPC now anyway. I'm talking about other people and companies using newer PowerPC designs.

 

[chrisjames61]

The PowerPC 601, 603, 604 and G3 and G4 and G5 were very good processors. The problem was that Intel and AMD cpu's of the day were "good enough" and cheaper. That and the economies of scale due to Windows being ubiquitous. Plus the clockspeed of the PPC cpu's languished for literally years. The G5 also wasn't suited for mobile use. The end came in 2006.

 

[SarahKerrigan]

I doubt Apple will go back to PPC now anyway. I'm talking about other people and companies using newer PowerPC designs.

Newer PowerPC designs are perfectly fine. I'm not sure why you think IBM's current-generation processors have some massive amount of "catching up" to do. P9 is starting to show its age but that's because it's midway through its product cycle (and got hit hard by Spectre mitigation), not because PPC is somehow far behind.

 

[chrisjames61]

Also chicken and the egg thing. Their is paltry if any software for the PowerPC as far as anything consumer related.

 

[DrMrLordX]

Newer PowerPC designs are perfectly fine. I'm not sure why you think IBM's current-generation processors have some massive amount of "catching up" to do. P9 is starting to show its age but that's because it's midway through its product cycle (and got hit hard by Spectre mitigation), not because PPC is somehow far behind.

When can we expect Talos III systems based on POWER10?

 

[SarahKerrigan]

When can we expect Talos III systems based on POWER10?

The public roadmap says Power10 is 2021.

 

[whm1974]

When can we expect Talos III systems based on POWER10?

I don't think that is considered to be consumer level hardware. Well Price wise it isn't.

 

[DrMrLordX]

The public roadmap says Power10 is 2021.

Hmm. Well it'll be interesting to see when/if Talos III systems emerge based on POWER10. I fear they'll be a bit behind the curve compared to competitors.

I don't think that is considered to be consumer level hardware. Well Price wise it isn't.

It really isn't. It's a fairly important piece of equipment for anyone that wants to profile code at home for possible deployment at work later. You wouldn't buy it for a browser box or a gaming rig. Even as a productivity machine, I'm afraid Talos systems likely fall short.