@olegil

Quote:

I recently evaluated the OMAP35xx family compared to the MPC5121e, and both TI and Freescale (and their common distributor in Norway, Arrow) agreed that the 400MHz PPC beat the 500MHz Cortex-A8 (they spec it to 600MHz in the data sheet, but that requires higher voltage and reduces life time to 1 year running. I needed an always on project). But the real clincher was that it has less features for more money, given the project needs. Honestly, the beagle board will not beat the SAM. Doesn't mean the beagle board isn't a very cool device (I do have the eval board, just wish I had some time to play with it), but it helps to get the facts straight.

Indeed, such as Sam does not use the MPC5121e

However even if Sam is faster, it's not going to be by a huge margin exactly.

Quote:

There are rumours about 1GHz Cortex-A8 processors, but none exist. The fastest you can actually go TODAY is 500MHz. Overclockable to 600MHz in extremely short lived applications.

Er, Beagleboard is 600MHz and has been for some time.
Snapdragon (similar to A8) is shipping at 1GHz in the Toshiba TG01, they've announced a 1.5GHz dual core version.
i.mx515 is 800MHz, don't know if they're shipping yet but it will be soon as there's devices with it due soon.

Anyway, what does it matter what's shipping TODAY? if there was to be an OS4 port to ARM it wont be anytime soon.

_________________
Whyzzat?

@BigD

slightly off topic

you can force some models of PPC mac to boot from a usb drive from the open firmware prompt. i forget what macs it works on but if i find the information i will pass it on

_________________

@TheMaskedMuchacho

Quote:

TheMaskedMuchacho wrote: from wikipedia Quote: PowerPC 750CL 750CL is an evolved 750CXe, with speeds ranging from 400 MHz to 1 GHz with a system bus up to 240 MHz, L2 cache prefetch features and graphics related instructions have been added to improve performance. As the added graphics-related functions match closely the ones found in the Gekko processor it is very likely that the 750CL is a shrink of the same processor for general purpose use. The 750CL is manufactured using a 90 nm copper based fabrication with Low-K dielectric and Silicon on insulator technology. It has 20 million transistors and the die is 16 mm² small. It draws up to 2.7 W at 600 MHz, 9.8 W at 1 GHz. Calling 729 MHz unsupported sounds a little bit silly to me. Saying no 729 MHz model exists would be better. Broadway This Broadway processor is manufactured in East Fishkill, NY, but it's packaged in IBM's plant in Canada, hence the text on the metal cover. It is believed but not confirmed that the processor in the Nintendo Wii game console is either a 750CL or a modified version of it. For instance, it runs at reported 729 MHz, a frequency not supported by stock 750CL. It measures only 4.2 x 4.5 mm (18.9 mm²). This is smaller than half the size of the "Gekko" microprocessor (43 mm²) incorporated in the GameCube at its first release. The chip may contain more instructions or higher SIMD support similar to the modifications made to the Gekko off the base 750CXe.

Could someone please explain to me why a chip that can be clocked up to 1 GHz would not be able to run at 729 MHz? Sure, it would be weird to underclock a chip to 729 MHz, but there could always be valid reasons for doing so (e.g., maybe it can be passively cooled at that speed).

Hans

Last edited by Hans on 21-Jul-2009 at 10:29 PM.

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Arko

Quote:

AFAIK there are already 2GHz ARMs around, but what is missing is a simple main-board, here are some AOS4 fans who tell Hyperion they should go to ARM, but they could not post a link to an existing system.

I haven't heard of any 2GHz ARM, but I'm not that dedicated either. IMO, they should go x86_64 but ARM is a viable alternative considering what is in the pipeline. Yeah, going to break many things if not everything but it gives them an excuse to break most things or not everything in pushing OS4 beyond the handcuffs of 3.1. Evolve or die.

Quote:

So why should Hyperion switch to ARM if there are no existing motherboards ? As long as there are only CPUs you can use them for decorative purpose, but not for an OS.

But there are mobos out there. Beagle and EFIKA-MX to name two and I'm not all that up on ARM mobos.

Quote:

Today it would make more sense if Hyperion would support some Mac PPC hardware, if they have done this, they will have enough time to port AOS4 somewhere else.

Probably, but they are still putting time and effort into a doomed Arch, that is not a smart move either to be wasting that valuable on a dead end.

Quote:

I know, but the board sold now, is not meant to be a consumer board, I don't blame Genesi for the price, but a Sam440 is cheaper for customers today.

Once OS4 is running on it, then we can have a fair comparison. Until OS4 is ported and Genesi goes into mass production of the EFIKA-MX, it's not viable to do a comparison between SAM440 and EFIKA-MX.

_________________
Dammy

arm is good for a PDA but too week for an desktop cpu

@Hans

It's running at 729 from a multiplier of 3, with FSB at 243MHz. This is unsupported by the 750CL (the datasheet specifies max 240).

So they're overclocking the memory interface (compared to the regular chip) and underclocking the core. Think of it as the reverse of the Celeron of old

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@minator

Hmm. Seems the numbers for OMAP35xx is 50000 hours at full speed, 100000 hours at 500MHz if 600MHz is used less than 23% of the time (datasheet page 123).

Still, for an industrial application this is not good enough. For a handheld device it is awesome. But it means the Beagle Board is either sold overclocked with reduced life span or it's got an adjustable voltage regulator for dynamic clock / voltage scaling. We didn't want either solution on our board

The i.MX31 had something like 1 year at full speed (because you needed a much higher voltage for that) or 10 years at normal speed.

There's almost 8800 hours in a year, in case you want to compare them



Edit:
And I know the SAM doesn't use the MPC5121e, but if the OMAP3530 at 500MHz isn't faster than a PPC at 400MHz, how on earth can it be faster than a PPC at 600MHz (and higher)?

Last edited by olegil on 22-Jul-2009 at 05:02 AM.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@Leo

Quote:

Btw I'm pretty 95% of Mac users wouldn't see a difference between an intel Mac and a PowerPC Mac...

The original Mac users would have. "PowerPC" was a catch phrase for the Mac. G5 represented power. And "AltiVec" was the superior 64-bit vector engine.

Perhaps MMX is better integrated into an Intel now with it's own registers, I don't know. But I still consider the PowerPC version which had it's own dedicated registers and vector excecution unit better than the orginal Intel hack that needed to dump the FPU registers on the stack because they were used as vectors and the only ones 64-bits wide.

Quote:

Linux is not better on PPC than on x86

Does that mean Linux is better on x86 than PPC? Infact it is; more specifically, it is better on Intel than AMD. And better without name brands such as NVidia where there are no generic drivers properly working out of the box. Having the latest closed source 3d hardware doesn't help with the open source nature of Linux.

Quote:

Just like AmigaOS isn't better on PPC than on 68k.

It is here! I can survive longer with my faster PPC compared to a slower 68k.

@TheMaskedMuchacho

Quote:

An A1200 in a tower case with a voodoo3 and a sb live meets all your criteria but can do so with OS3.1, the only thing os4 does is remove the dependency the on custom chips and this could have been done without a move to power pc. obviously there is a lot more to os4 than removing dependencies but thats all thats relevant in this context.

I did realise this when posting.

Yes, the PCI cards did bring this about first. Almost a pitty, when Amiga hardware was the superior at graphics and audio, and then we start using expansion cards so we can plug in PC graphic and sound cards. Did we lose it then?

Only Amiga people would purchase an expensive expansion card in order to plug other cheapo cards into it! Logically that doesn't make sense.

But I might disagree with you on one point, about the tower. Only if the A1200 has been easily slotted into a generic PC case can it meet the criteria. If it is a special A1200 case then that looses out as they were another expensive add on.

@BigD

Quote:

1) In PPC favour - You can run the Mac Classic version of Settlers 2 on a PPC Mac but not an Intel Mac!

You can also run Payback, A splitting image of the Amiga version!

Oh, doesn't Rosetta work? I never saw it working on a MacOSX86 hack. But on a "real" x86 Mac whitebox it should work.

Quote:

2) In favour of Intel - Intel Macs can boot off an external USB hardrive

Not in favour of Apple. Thay scrapped their ow nformware and really turned the Mac into a PC. With a disc stuck in the DVD drive you will see a black screen with white text looking exactly like a PC telling you to insert a bootable disc in the drive and press the any key. Now, Apple may have ripped off the customers by using cheap Intel chips and still charging Apple prices. But practically just sticking in a PC mainboard, BIOS and all, into a Mac whitebox and calling that a Mac!? No way! That's an insult! Apple heads should be ripped of for that!

Safari is released on Windows. So when is OSX going to be released on Windows PC? It already runs on the hardware. Won't be long before the Mac suffers "Amiga" pressure and the only identifiable thing left about the Mac will be an OS.

@Hypex

Quote:

Hypex wrote: Yes, the PCI cards did bring this about first. Almost a pitty, when Amiga hardware was the superior at graphics and audio, and then we start using expansion cards so we can plug in PC graphic and sound cards. Did we lose it then?

Err...no. By the time commodity PCI sound and graphic cards were available for the Amiga, the Amiga's native chipset had long since been eclipsed.

_________________
The problem with AmigaOS on PPC isn't that PPC is big-endian. The problem with AmigaOS on PPC is that PPC is dead-endian.

@olegil

Hmm. Went to check the numbers on the i.MX51, and can't find the data sheet.

Freescale has been hit (along with Qualcomm, Freescale, ST Microelectronics, ATI Technologies and Spansion) with an ITC ruling that prevents them from selling certain chips to the US market. Seems kinda bad. The ruling became effective as of yesterday.

Qualcomm (who makes SnapDragon) has ALSO been hit with a ruling in a patent case against Broadcom. Specifically on 6 patents regarding 3G functionality (which is the majority, if not the whole, of the market for their ARM chips).

The fact that they're fighting rather than paying royalties indicates that they think they have a case, of course.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

Quote:

olegil wrote: @Hans It's running at 729 from a multiplier of 3, with FSB at 243MHz. This is unsupported by the 750CL (the datasheet specifies max 240). So they're overclocking the memory interface (compared to the regular chip) and underclocking the core. Think of it as the reverse of the Celeron of old

Now that makes sense, although I do wonder if an extra 3 MHz for the FSB would be worth the effort.

Hans

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

Quite possibly there is some timing effect involved. Like the Amiga vs Atari thingie where the Amiga clocked everything in sync with the PAL or NTSC master clock because of the extremely tricky DMA timing between chips on the board while the ST just had a CPU clocked at 8MHz and a video chip working on PAL/NTSC reference.

The Wii is much closer to the Amiga than the ST in that regard.

Here you can see a claim that the GPU also runs at 243MHz:
http://en.wikipedia.org/wiki/Wii#Technical_specifications

With very little GPU memory (just 3MB) it makes sense to tune both chips to the same FSB clock and do some trickery on the memory interface. Possibly the main CPU is leeching bandwidth off of the GPU memory controller or something cool like that. I don't think Broadway has an internal memory controller itself. But I could be wrong. It's been known to happen

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

i.MX has 1.25 years lifetime at full speed, 10 years at lower speed.

This is just so bad there's no end to how bad that is. For anything except a handheld battery operated device with limited useability anyway, that is.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@Hypex

Quote:

With a disc stuck in the DVD drive you will see a black screen with white text looking exactly like a PC telling you to insert a bootable disc in the drive and press the any key.

Could you expand on what Intel Mac you have seen such behaviour with?

Because between the various Mac laptops and the Mac Minis which I have had the pleasure of controlling, not one of them will even try to boot off the CD/DVD unless you specifically ask it to. Which, in the case of the minis, is rather annoying for us --- means we need to have a keyboard around when first installing them.

I still hope it's a PS3 port.

my guess:

It's a Bill McEwan vodoo doll.

_________________
Bite Me!

@brotsalami

Quote:

It's a Bill McEwan vodoo doll.

And believe it or not: It's on schedule and rockin'!

OldFart

_________________
More then three levels of indigestion and you're scroomed!

@umisef

He's indeed making things up, while he's right Apple don't use their own implementation of OpenFirmware anylonger but EFI instead.
There is no support for VGA/VESA/Textmodes etc in Apples EFI, DVDs/CDs are indeed ignored if insert unless you explicitly select it in the boot chooser (holding down option/alt when power on) or hold down the C or D key.
You can however atleast on PowerPC macs set the boot device to the CD drive and thus- if a disk is insert it will automatically boot it, but it won't ask you about it.