@pavlor

Quote:

pavlor wrote: @cdimauro Quote: I simply recognized a logical fallacy, and "raised the flag". That's all. Quote: Maybe because I did it several times in the last days, so it comes naturally now. Again about your "logic"?

About the definition of "Ad hominem fallacy", which isn't mine.
Quote:

Quote: e6500 is the only processor that makes sense to use. e5500 is too much limited, especially at that price range. e5500 is first SoC I could imagine for such price.

Without Altivec...

@cdimauro

Quote:

About the definition of "Ad hominem fallacy", which isn't mine.

It shows, your usage of this term is quite excessive.

Quote:

Without Altivec...

You mean like other hardware available for OS4 (440EP/460EX, 750CXe/FX/GX)? That should be no problem for me.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: About the definition of "Ad hominem fallacy", which isn't mine. It shows, your usage of this term is quite excessive.

I only care if the such definition applies or not.
Quote:

Quote: Without Altivec... You mean like other hardware available for OS4 (440EP/460EX, 750CXe/FX/GX)? That should be no problem for me.

Yes, I have the same position.

@cdimauro

Quote:

I only care if the such definition applies or not.

I see, if it suits your needs... whatever they are.

@pavlor: is it wrong? I think that it's much better than someone which cherry-pick information to create a different History...

@cdimauro

Quote:

is it wrong?

From you? I would be worried if it would be other way.

@pavlor: the judge talked...

@cdimauro

Quote:

@pavlor: the judge talked...

Then abide by his ruling.

Had the opportunity to meet Trevor once more, this time I handed out sheets all about my past projects and the big one I am working on right now - Evolve. After speaking with Christopher Follett @ AmigaKit I have decided it would be a great idea to add support for ProAction - although that may come a little later.

http://myevolve.wordpress.com

Last edited by djrikki on 29-Aug-2015 at 01:00 PM.

_________________

@cdimauro

@cdimauro

Quote:

cdimauro wrote: @pavlor Quote: pavlor wrote: @cdimauro [quote]I simply recognized a logical fallacy, and "raised the flag". That's all. Quote: Maybe because I did it several times in the last days, so it comes naturally now. Again about your "logic"?

About the definition of "Ad hominem fallacy", which isn't mine.
Quote:

Quote: e6500 is the only processor that makes sense to use. e5500 is too much limited, especially at that price range. e5500 is first SoC I could imagine for such price.

Without Altivec...[/quote]

A revised and improved version of AltiVec is due to be reintroduced in later revisions of Freeescale's PPC lines, in the case of the higher end e5500 cored processors when the P prefix processors are revised to T prefix processors.
AND the processors that Acube uses don't have AltiVec instructions either. While they are useful, its not that big a loss.

The processor in the X1000 has AltiVec instructions, but the X5000 is still going to run circles around it.
AND it should be possible for a latter revision of the X5000 to use the AltiVec enhanced cpus.

@iggy

Quote:

AND it should be possible for a latter revision of the X5000 to use the AltiVec enhanced cpus.

Only if Freescale releases suitable SoC. P5020 in X5000 is pin compatible with P5010, P5040, P4080 and P3041. T1042 is pin compatible with T2081 (which is e6500 core based).

@pavlor

No word yet on pin compatibility as transition from P series to T series occurs, but it seems likely.
And even without AltiVec, I find the X5000 appealing.
Also, as a big fan of the T1024, I'm sure T10XX based boards are inevitable.
Finally, on the e6500 core, this IS what we need for a final PPC migration.
It is shown in Freescale's longest term roadmaps and is likely to still be in production for some time now (my guess would be at least ten years).

@iggy

Quote:

iggy wrote: @cdimauro Quote: cdimauro wrote: @pavlor Without Altivec... A revised and improved version of AltiVec is due to be reintroduced in later revisions of Freeescale's PPC lines, in the case of the higher end e5500 cored processors when the P prefix processors are revised to T prefix processors.

Are you sure that it'll be introduced with the e5500 family? It's the e6500 which supports Altivec. Do you know something else?
Quote:

AND the processors that Acube uses don't have AltiVec instructions either. While they are useful, its not that big a loss. Some time ago an OS4 developer said that Altivec is supported and used. Also, applications compiled for it have shown a good performance boost. [quote]The processor in the X1000 has AltiVec instructions, but the X5000 is still going to run circles around it.

The X1000's CPU was a disappointment, but in a single core/thread scenario it should perform similarly to an e5500 or e6500. So, were you talking about a multicore scenario? I mean, aggregate results (of all cores and/or threads)?
Quote:

AND it should be possible for a latter revision of the X5000 to use the AltiVec enhanced cpus.

The X5000 should use the e5500, which has no Altivec. Have you any new about it?

@cdimauro

Quote:

Are you sure that it'll be introduced with the e5500 family? It's the e6500 which supports Altivec. Do you know something else?

I think he meant Freescale will introduce TXXX equivalent of P50X0.

Quote:

but in a single core/thread scenario it should perform similarly to an e5500 or e6500.

Coremark result of e5500 is noticeably higher than PA6T. However, today we know only one X5000 benchmark result (Cow3D), there X5000 leads over X1000.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: Are you sure that it'll be introduced with the e5500 family? It's the e6500 which supports Altivec. Do you know something else? I think he meant Freescale will introduce TXXX equivalent of P50X0.

OK, but the T1 series uses an e5500 instead of the e6500 used by the T2 and T4 ones.
Quote:

Quote: but in a single core/thread scenario it should perform similarly to an e5500 or e6500. Coremark result of e5500 is noticeably higher than PA6T.

Do you have some data for both? Single core/thread, of course.
Quote:

However, today we know only one X5000 benchmark result (Cow3D), there X5000 leads over X1000.

With the same graphic card?

@cdimauro

Quote:

Quote: However, today we know only one X5000 benchmark result (Cow3D), there X5000 leads over X1000. With the same graphic card?

Yes. The X5000 performaned substantially better than the highest benched X1000 setup.

I've seen the X5000 in action at various shows and meets this year and in general use it appears to be much faster than the X1000.

It's supposed to go on sale next month and once it's in the hands end users we can expect to see every aspect of the hardware tested. I fully expect it to beat X1000 in pretty much every area and wouldn't be surprised if performs better with Mplayer despite not having SIMD support.

@cdimauro

Quote:

Do you have some data for both? Single core/thread, of course.

Freescale has Coremark scores for T10XX SoC(s):
1400 MHz: 5180 (single core)

Coremark score for PA6T is no longer available, but it was cca 2750 for PA6T 1.7 GHz (single core, so cca 3000 for 1.8 GHz)

Note, I don´t think these scores are directly comparable (not same compiler settings etc.). You may better use e500v2 1.2 GHz scoring 3720 (single core) mentioned by Freescale in comparison tables.

Quote:

With the same graphic card?

Of course.

@Rob

Quote:

Quote: Rob wrote: @cdimauro With the same graphic card? Yes. The X5000 performaned substantially better than the highest benched X1000 setup. I've seen the X5000 in action at various shows and meets this year and in general use it appears to be much faster than the X1000. It's supposed to go on sale next month and once it's in the hands end users we can expect to see every aspect of the hardware tested. I fully expect it to beat X1000 in pretty much every area and wouldn't be surprised if performs better with Mplayer despite not having SIMD support.

Fine. However other applications that make used of Altivec might show the contrary.

Last edited by cdimauro on 29-Aug-2015 at 06:24 PM.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: Do you have some data for both? Single core/thread, of course. Freescale has Coremark scores for T10XX SoC(s): 1400 MHz: 5180 (single core) Coremark score for PA6T is no longer available, but it was cca 2750 for PA6T 1.7 GHz (single core, so cca 3000 for 1.8 GHz) Note, I don´t think these scores are directly comparable (not same compiler settings etc.). You may better use e500v2 1.2 GHz scoring 3720 (single core) mentioned by Freescale in comparison tables.

That's a huge different, indeed.

I'm looking forward to see some real-world application.