Return to Level1Techs.com

The POWER and PowerPC General Discussion / News Thread

news
mega_thread
architecture

#61

Spectre vulnerability was expected, but Meltdown (CVE-2017-5754) I thought was limited to Intel and some ARM. Maybe I need to add POWER to that list.


#62

There’s no mincing words about it, they’re vulnerable to both.


#63

If I had some spare cash I’d be tempted to pick this up for £250, mosty for nostalga, but it could run MorphOS I think:


#64

It could, but the 450MHZ in that thing always wants to be as hot as possible. Not wurf.


#65

I have an original phat PS3 running debian


#66

its a VM :stuck_out_tongue: need an E3 chip to actually install linux.


#67

True but I am running custom firmware and running all 8 SPEs with GameOS access to the hardware. Most of the HV is out of the way.


#68

I’m extremely jealous.


#69

Yeah but I cannot get a modern kernel on the damn thing. The 3.16 ABI break prevents kexec from loading a new kernel.

I was wondering if dropping 50USD on an E3 and another 50 on the BDROM are worth it. I am a PCC fanboy but, 256MB of real system RAM…


#70

Drop an SSD in there and use the swap for ram, swap the system ram over to the GPU, play halo.


#71

I was using the GPU RAM as swap. The issue was that I was working on the nouveau driver port for PS3 before the ABI break. I would like to pick that up again so that there is a fully native driver with full hardware acceleration. Can’t do that when using the GPU memory.
#firstworldproblems


#72

That and its only the 304.11 driver :T


#73

Well, I got a reply back from Raptor, in which they state:

In a nutshell, only the DD2.2 silicon changes were needed. DD2.2 silicon is able to close off these security holes, with the exception of the Spectre same-process read vulnerability that affects the entire CPU industry, with only changes to firmware and a small kernel change. Note that the kernel change is minimally invasive and is not related to the 30% performance loss Intel KPTI mitigation.

It would probably be more helpful if I understood production stepping better.


#74

From @wuttztfz’s post on the immune CPUs list.


#75

Thank your for this update. Your efforts are appreciated :hugs:

I’m removing Power6 and adding PPC750 and PPC7400


#76

I pulled this info from your thread though :confused:; what did you find that disqualifies POWER6? The Wikipedia page says that it worked in-order, I thought out-of-order was needed to allow these exploits.


#77

He tested it successfully on Power6


#78

Wait ok is the 750, 7400, 7445, 7450/e, and 7600 chips under a general POWER denotation? I’ve always wondered if they were but never have really looked it up.


#79

No, since POWER refers to the line of IBM processors, which these were not since they were PowerPC. You could call them Power processors, as that would imply the general term Power Architecture:

Writing POWER rather than Power becomes really confusing, really fast because there is an old IBM POWER ISA which is different from the modern Power ISA.

It is easiest to look at the history of the POWER chips:

IBM POWER ISA PowerPC ISA Power ISA
POWER1 → POWER2 POWER3 → POWER5 POWER6 → POWER9

It looks like many PowerPC chips are compliant with Power ISA, even if they were made before Power ISA was created. So you would probably be correct in calling them Power but not POWER since that would suggest either the chip line or the old ISA, neither of which applies.


#80

So more like POWER-1 :stuck_out_tongue: personally, at this point, I want a PS3 hacked with freebsd and lumina and I’ll be happy.