New SAM460EX
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 4977 from 2009/1/28
    From: Delaware, USA
    You got far closer to the truth at a distance than I did via direct inquiry, Andreas. And, yes, that sounds about the right date. Further, not only did their reps claim that their partners were testing the first Titan based processors. They claimed this was the first of a family of processors. The reasoning (as presented to me) for not accepting my request for an NDA is that they preferred to work (initially) with the small group of qualified partners they were already working with.
    Having just finishing a couple of months of inquiries to IBM, that logic sounded a lot like IBM logic (which made sense considering AMCC's origins).
    Now you tie together all these statements that don't agree with each other. What other conclusion is there other than that Applied Micro concealed that it was not moving forward with the long touted Titan core and was instead moving to a backup plan?
    The problem is I can't think of an alternate reason for these circumstances. I don't lean to paranoid speculation, but I do feel deceived. And I don't really trust the company.
    "Never attribute to malice what can more readily explained by incompetence"
  • »04.10.10 - 01:12
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > that sounds about the right date.

    You mean that Applied Micro rejected your NDA request just before you told me here on MorphZone, i.e. in late June or very early July?

    > not only did their reps claim that their partners were testing the first
    > Titan based processors.

    Which I believe was the truth, judging from all information we have to date.

    > The reasoning (as presented to me) for not accepting my request for an NDA
    > is that they preferred to work (initially) with the small group of qualified partners
    > they were already working with.

    That sounds somewhat reasonable. But didn't you say before that it was the odd requirement to maintain a website?

    > that logic sounded a lot like IBM logic (which made sense considering AMCC's origins).

    AMCC was founded in 1979. They acquired IBM's PPC4xx processor business (including some 70 engineers, but no management people) only in 2004, which I think is much too little connection between the two companies to draw such a conclusion. So I think this same pattern you experienced is rather coincidence.

    > you tie together all these statements that don't agree with each other.

    I'd rather say I contrast them with each other.

    > What other conclusion is there other than that Applied Micro concealed
    > that it was not moving forward with the long touted Titan core and was
    > instead moving to a backup plan?

    Whatever answer I'd give to this question it wouldn't explain the Linley Group even adding to the confusion by making claims (that Titan and the first CPU to be based on it were scrapped *before* that CPU was publically announced) that contradict both Applied Micro's story as presented to the public and what probably really happened. I don't think that the Linley Group's recent claim, which even challenges an older statement from them (presenting Apple's Intrinsity acquirement in April 2010 as reason), stands unchallenged.

    > I do feel deceived.

    Now imagine how their investors and share holders will feel as soon as they find out (or did they already?) ;-)
  • »04.10.10 - 04:02
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Andreas, Jim, could you post some kind of resume, or a status report about this fascinating discussion? It's not easy to follow, but right not, mi conclusion is that freescale is the best (read: less worse) PowerPC chip provider (ugh!)

    Could you add the IBM PowerEN to your discussion? It was announced back on february, and I don't recall seeing it mentioned here:

    Quote:

    Roberto Innocenti wrote on powerdeveloper.org:

    A2 Core, full 64 bit based on Power Architecture? technology and support Power ISA 2.06
    2.3GHz 45nm SOI with 16 cores and 64 Threads
    Low power, highly scalable design - 25-75W chip at full frequency


    Alright, it's a router/server thingie, but very interesting, don't you think?

    [ Edited by jcmarcos on 2010/10/5 17:26 ]
  • »05.10.10 - 06:42
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > Could you add the IBM PowerEN to your discussion?
    > [...] I don't recall seeing it mentioned here

    Huh? Are there several 'jcmarcos' on this forum? ;-) That other 'jcmarcos' even quoted me and my link discussing PowerEN less than 2 months ago:

    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7001&forum=3&post_id=75802#75802

    > It was announced back on february

    Yes, but not yet dubbed "PowerEN" back then:

    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6268&forum=11&post_id=71155#71155

    Btw, I think that Roberto Innocenti posted to the wrong thread because, opposed to what he seems to imply, the PowerEN has nothing to do with his fantasy "cell quad-core processor" for PSP2 which he started the thread about. He should have posted his recent comment there:
    http://www.powerdeveloper.org/forums/viewtopic.php?t=1847
  • »05.10.10 - 18:50
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Andreas_Wolf wrote:
    Quote:

    jcmarcos wrote:
    Could you add the IBM PowerEN to your discussion? I don't recall seeing it mentioned here


    You even quoted me and my link discussing PowerEN less than 2 months ago


    Whoooops! Sorry... That's why I said that I didn't recall. Anyway, what do you think about the PowerEN? Or it's lack of corporate drivel, like the Titan, makes it uninteresting? :-D
  • »06.10.10 - 14:42
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > what do you think about the PowerEN?

    In the context of mobile (read: netbooks/notebooks) or desktop computing I don't find it that much interesting because it seems to lack in on-chip peripheral controllers: there's apparently just (as much as four) 10GbE and only two PCIe (Gen2) lanes, nothing more. See this presentation from August/September:

    http://www.power.org/events/POWERWEBINAR082010/IBM_final_powerorg_2010Aug31_-_Gene.pdf
  • »06.10.10 - 15:28
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Andreas_Wolf wrote:

    I don't find it that much interesting because it seems to lack in on-chip peripheral controllers


    Sure. It's no more interesting than any other router processor. Pity. So back to the one-and-only QorIQ speculation, I guess...
  • »07.10.10 - 07:35
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > back to the one-and-only QorIQ speculation, I guess...

    QorIQ, while definitely being the most interesting one due to the recent AltiVec announcement, is not the only current/future Power Architecture processor that has more than Ethernet and PCIe: PacketPro has USB2/3, SATA and SDHC, and Axxia has USB2 and something denoted "Flash" in the block diagram (which probably means some SD standard).
  • »07.10.10 - 13:22
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > the powermac is much better

    Comparison Sam460ex at 1.167 GHz (OS4) vs. Mac mini G4 at 1.5 GHz (MorphOS):

    Sam460ex results:

    Code:
    ---> CPU <---
    MAX MIPS: 2332

    ---> L1 <---
    READ32: 4511 MB/Sec
    READ64: 9011 MB/Sec
    WRITE32: 4449 MB/Sec
    WRITE64: 8882 MB/Sec

    ---> L2 <---
    READ32: 1061 MB/Sec
    READ64: 1061 MB/Sec
    WRITE32: 521 MB/Sec
    WRITE64: 520 MB/Sec

    ---> RAM <---
    READ32: 311 MB/Sec
    READ64: 310 MB/Sec
    WRITE32: 521 MB/Sec
    WRITE64: 521 MB/Sec
    WRITE: 1251 MB/Sec (Tricky)

    ---> VIDEO BUS <---
    READ: 72 MB/Sec
    WRITE: 261 MB/Sec

    Source: http://www.amigans.net/modules/newbb/viewtopic.php?post_id=59270#forumpost59270


    Mac mini G4 results (my machine):

    Code:
    ---> CPU <---
    MAX MIPS: 4475

    ---> L1 <---
    READ32: 5687 MB/Sec
    READ64: 11372 MB/Sec
    WRITE32: 4874 MB/Sec
    WRITE64: 3794 MB/Sec

    ---> L2 <---
    READ32: 2579 MB/Sec
    READ64: 3363 MB/Sec
    WRITE32: 2330 MB/Sec
    WRITE64: 2952 MB/Sec

    ---> RAM <---
    READ32: 387 MB/Sec
    READ64: 403 MB/Sec
    WRITE32: 771 MB/Sec
    WRITE64: 771 MB/Sec
    WRITE: 809 MB/Sec (Tricky)

    ---> VIDEO BUS <---
    READ: 32 MB/Sec
    WRITE: 180 MB/Sec


    Used benchmarking program is RAGEMEM v0.37 (run on MorphOS via OS4Emu):
    http://os4depot.net/?function=showfile&file=utility/benchmark/ragemem.lha
  • »11.11.10 - 23:38
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    Update:

    > PPC470: 2.3 or 2.5 (varying with information source)

    According to IBM* the PPC476FP core even delivers 2.7 DMIPS/MHz. That puts it above e500mc (2.5 DMIPS/MHz).

    * PPC476FP Core Product Brief (August 2010, page 2)
  • »12.11.10 - 23:55
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > Comparison Sam460ex at 1.167 GHz (OS4) vs. Mac mini G4 at 1.5 GHz (MorphOS):
    > [...]
    > Used benchmarking program is RAGEMEM v0.37 (run on MorphOS via OS4Emu)

    Comparison Sam460ex at 1.0 GHz (OS4) vs. Mac mini G4 at 1.5 GHz (MorphOS):

    Sam460ex results:

    Code:
    [Dec 12 14:52:26 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar).
    [Dec 12 14:52:45 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar)
    0.00:00:17.07 [10,124,948 nodes/sec]
    [Dec 12 14:52:45 UTC] OGR-NG benchmark summary :
    Default core : #-1 (undefined)
    Fastest core : #0 (KOGE 3.1 Scalar)
    [Dec 12 14:52:45 UTC] RC5-72: using core #0 (MH 2-pipe).
    [Dec 12 14:53:04 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe)
    0.00:00:16.44 [3,055,014 keys/sec]
    [Dec 12 14:53:04 UTC] RC5-72: using core #1 (KKS 2-pipe).
    [Dec 12 14:53:22 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe)
    0.00:00:16.08 [3,286,052 keys/sec]
    [Dec 12 14:53:22 UTC] RC5-72: using core #2 (KKS 604e).
    [Dec 12 14:53:40 UTC] RC5-72: Benchmark for core #2 (KKS 604e)
    0.00:00:16.07 [3,137,568 keys/sec]
    [Dec 12 14:53:40 UTC] RC5-72: using core #5 (MH 1-pipe).
    [Dec 12 14:53:59 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe)
    0.00:00:16.08 [2,964,053 keys/sec]
    [Dec 12 14:53:59 UTC] RC5-72: using core #6 (MH 1-pipe 604e).
    [Dec 12 14:54:18 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e)
    0.00:00:16.07 [2,981,904 keys/sec]
    [Dec 12 14:54:18 UTC] RC5-72 benchmark summary :
    Default core : #-1 (undefined)
    Fastest core : #1 (KKS 2-pipe)

    Source: http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=32858&forum=33#591840


    Mac mini G4 results (my machine):

    Code:
    [Dec 12 19:51:51 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar).
    [Dec 12 19:52:11 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar)
    0.00:00:16.62 [14,881,017 nodes/sec]
    [Dec 12 19:52:11 UTC] OGR-NG: using core #1 (KOGE 3.1 Hybrid).
    [Dec 12 19:52:30 UTC] OGR-NG: Benchmark for core #1 (KOGE 3.1 Hybrid)
    0.00:00:17.23 [31,267,175 nodes/sec]
    [Dec 12 19:52:30 UTC] OGR-NG benchmark summary :
    Default core : #1 (KOGE 3.1 Hybrid)
    Fastest core : #1 (KOGE 3.1 Hybrid)
    [Dec 12 19:52:30 UTC] RC5-72: using core #0 (MH 2-pipe).
    [Dec 12 19:52:49 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe)
    0.00:00:16.11 [5,763,995 keys/sec]
    [Dec 12 19:52:49 UTC] RC5-72: using core #1 (KKS 2-pipe).
    [Dec 12 19:53:08 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe)
    0.00:00:16.99 [5,647,545 keys/sec]
    [Dec 12 19:53:08 UTC] RC5-72: using core #2 (KKS 604e).
    [Dec 12 19:53:27 UTC] RC5-72: Benchmark for core #2 (KKS 604e)
    0.00:00:16.63 [5,773,289 keys/sec]
    [Dec 12 19:53:27 UTC] RC5-72: using core #3 (KKS 7400).
    [Dec 12 19:53:46 UTC] RC5-72: Benchmark for core #3 (KKS 7400)
    0.00:00:16.09 [13,421,897 keys/sec]
    [Dec 12 19:53:46 UTC] RC5-72: using core #4 (KKS 7450).
    [Dec 12 19:54:05 UTC] RC5-72: Benchmark for core #4 (KKS 7450)
    0.00:00:16.80 [15,701,333 keys/sec]
    [Dec 12 19:54:05 UTC] RC5-72: using core #5 (MH 1-pipe).
    [Dec 12 19:54:24 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe)
    0.00:00:16.70 [5,261,511 keys/sec]
    [Dec 12 19:54:24 UTC] RC5-72: using core #6 (MH 1-pipe 604e).
    [Dec 12 19:54:43 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e)
    0.00:00:16.10 [5,198,168 keys/sec]
    [Dec 12 19:54:43 UTC] RC5-72 benchmark summary :
    Default core : #4 (KKS 7450)
    Fastest core : #4 (KKS 7450)


    Used benchmarking program is dnetc (v2.9106-514 on MorphOS and v2.9105-511 on OS4*), run as "dnetc -bench":
    http://www.distributed.net/Download_clients

    Clock per clock comparison of the CPUs regarding OGR-NG and RC5-72:
    In OGR-NG the 7447A/B is about 105% faster per clock than the 460EX. In RC5-72 the 7447A/B is about 220% faster per clock than the 460EX. This vast advantage of the 7447A/B is due to its ability to run AltiVec accelerated dnetc cores.

    Clock per clock comparison of the CPUs regarding the particular dnetc cores**:
    KOGE 3.1 Scalar: 460EX is 2% faster per clock.
    MH 2-pipe: 7447A/B is 26% faster per clock.
    KKS 2-pipe: 7447A/B is 15% faster per clock.
    KKS 604e: 7447A/B is 23% faster per clock.
    MH 1-pipe: 7447A/B is 18% faster per clock.
    MH 1-pipe 604e: 7447A/B is 16% faster per clock.

    * Using the very same version on MorphOS as on OS4 doesn't result in any significant changes.
    ** KOGE 3.1 Hybrid, KKS 7400 and KKS 7450 are N/A on 460EX due to missing AltiVec support in that processor.
  • »12.12.10 - 19:11
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Andreas_Wolf wrote:

    Comparison Sam460ex at 1.167 GHz (OS4) vs. Mac mini G4 at 1.5 GHz (MorphOS):

    Sam460ex results:

    ---> VIDEO BUS <---
    READ: 72 MB/Sec
    WRITE: 261 MB/Sec

    Mac mini G4 results (my machine):

    ---> VIDEO BUS <---
    READ: 32 MB/Sec
    WRITE: 180 MB/Sec


    Those figures surprise me. I never thought a SAM would be in any way trouncing a MacMini (or almost any other PowerPC AmigaOS/MorphOS computer). Could anyone ellaborate? What's "wrong" on the MacMini?
  • »13.12.10 - 09:10
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > What's "wrong" on the MacMini?

    While I don't have an answer to that question I just want to add that it's not only the Mac mini showing such low figures compared to the Sam460ex:

    ------------------------------
    Powermac G4 Gigabit (7455 1GHz)
    [...]
    VIDEO BUS
    READ: 38 MB/Sec
    WRITE: 126 MB/Sec
    ------------------------------
    http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=32673&forum=33#588009

    ------------------------------
    PowerBook 1.67GHz
    [...]
    --- VIDEO BUS ---
    READ: 32 MB/Sec
    WRITE: 159 MB/Sec
    ------------------------------
    http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=32673&forum=33&start=40#588294


    Pegasos II G4 seems to be placed between G4 Macs and Sam460ex:

    ------------------------------
    Pegasos II 1GHz
    [...]
    --- VIDEO BUS ---
    READ: 52 MB/Sec
    WRITE: 220 MB/Sec
    ------------------------------
    http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=32673&forum=33&start=40#588294


    For me it's not clear whether the Sam460ex results are derived from PCI (onboard GPU or Radeon in PCI slot) or from PCIe.
  • »13.12.10 - 09:51
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Zylesea
    Posts: 2050 from 2003/6/4
    There's nothing wrong wth the mini. But the 7447 and teh Apple northbridge aren't really a new chips. While the e600 ALU is nice and better than most other ppc ALUs, the 7447 uses a very old FSB interface and a dedicated northbridge.
    The Sam's 460 while not having a fast/high clocked ALU has a better FSB though (no wonder, the chip is low end but rather modern).
    Throughput is a major key factor for the overall performance. It is also the reason why an 8610 board would have outperformed all other G4 ppc systems on most operations (not pure numbercrunchuing like OGR though) - an e600 core with a fast FSB.
    Anyway, while the Sam460 figures show that this Sam will proabably offer at least a power nice enough for much it is also clear, that it is stil an entire class below the mini in terms of computing power.
    --
    http://via.bckrs.de

    Whenever you're sad just remember the world is 4.543 billion years old and you somehow managed to exist at the same time as David Bowie.
    ...and Matthias , my friend - RIP
  • »13.12.10 - 09:59
    Profile Visit Website
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Quote:

    Zylesea wrote:

    an 8610 board would have outperformed all other G4 ppc systems on most operations (not pure numbercrunchuing like OGR though) - an e600 core with a fast FSB.


    ...until you use its internal display unit, which steals CPU cycles the same way an Amiga does when using its custom chips. Or has it some mechanism for the DIU to read video mapped memory without blocking the CPU?

    Sorry to bring up again this old debate, but my curiosity on this has never been satisfied.

    Ulrich, didn't you actually own an 8610 evaluation board? Or was it Konstantinos?
  • »13.12.10 - 10:50
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > Ulrich, didn't you actually own an 8610 evaluation board? Or was it Konstantinos?

    I think it was the latter:

    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6196&forum=11&post_id=63178#63178
    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6465&forum=11&post_id=67222#67222
  • »13.12.10 - 14:13
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Zylesea
    Posts: 2050 from 2003/6/4
    Andreas is, no big surprise, right - I don't have an 8610. But I always liked to have seen one emerging on the market and would have suggested Acube to base their next Sam on it. Would have been the way smarter choice IMHO.

    The question whether the internal DiU is blocking the cpu or not is far from trivial. best choice is a 8610 board plus a (powerful) pci express gfx card with dedicated VRAM. But the DIU isn't as braking as the custom chips on Amiga were. The RAM controller runs at full speed (up to 533 MHz). The the maximum bandwith is shared between all DMA units attached to the bus. Since the bus is fast I woud expect a better performance with shared memory on a 8610 than with a slow bus and dedicated RAM on a 7447. Anyway, for high performance you better add a dedicated pci express gfx card and then the problem is void anyway.
    If you want 3D you'll need a dedicated fgx card anyway. But on youtube there is that demo which shows that the 8610 is cacpable of hd video - I think that shows quite well, that it is actually rather capable.
    Anyway - it's all academic. The 8610 is ageing. Today and tomorrow QorIQ the best route for PPC (if at all that is).
    --
    http://via.bckrs.de

    Whenever you're sad just remember the world is 4.543 billion years old and you somehow managed to exist at the same time as David Bowie.
    ...and Matthias , my friend - RIP
  • »13.12.10 - 15:05
    Profile Visit Website
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    To answer my own questions:

    > 1. Does PacketPro have an FPU?
    > [...]
    > 6. What is Applied Micro's "APM86XXX" "32-bit Commercial Processor" that is
    > listed in Power.org's latest Power Architecture roadmap version? Is it PacketPro?

    The answer is "yes" to both questions.

    "AppliedMicro's Mamba APM86190 single-core devices and APM86290 dual-core processors feature up to two 1.5GHz PowerPC 465 processing cores with floating point units, 32 KB I- and 32KB D-cache, 256 KB L2 cache per processor, hardware cache coherency, 1600 Mbps DDR3 memory controller with optional ECC. High-speed interfaces consist of two GE ports with classification and TCP/IP offload, one x4 PCI-Express(R) Gen2, two x1 PCI-e Gen 2 ports, two USB 2.0 hosts with integrated PHYs, one USB 2.0 OTG with integrated PHY and two SATA 2.0 ports. [...] Sample quantities of AppliedMicro's Mamba APM86190 single-core and APM86290 dual-core SoCs are available now and production quantities are expected in the second quarter of 2011."
    http://investor.appliedmicro.com/phoenix.zhtml?c=78121&p=irol-newsArticle&ID=1506830

    "Dual-Core Power 465 processors each with a Floating Point Unit"
    http://www.appliedmicro.com/products/process.html#EmbeddedProcessors

    "At the heart of the APM86290 are two 1.5-GHz 465 processor cores based on Power Architecture with full SMP support and individual Floating Point processors. [...] The APM86290 incorporates two high performance 465 processors. Each 465 has [...] an IEEE floating point unit (FPU)."
    http://www.appliedmicro.com/products/APM86290_PB_20101011.pdf

    I guess that both USB 3.0 and 10GbE, which were announced for PacketPro back in September, are supposed to be implemented only in future, yet to be announced PacketPro processors.

    Edit:
    http://www.apm.com/products/embedded/multicore460/apm86190/
    http://www.apm.com/products/embedded/multicore460/apm86290/

    [ Edited by Andreas_Wolf on 2011/1/6 22:48 ]
  • »13.12.10 - 19:34
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Andreas, that "mamba" chip looks the nicest PowerPC offer in ages (sans AltiVec, pity). As AMC is already sampling, could a development board already exist?
  • »14.12.10 - 07:05
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > that "mamba" chip looks the nicest PowerPC offer in ages

    At least the best Applied Micro has to offer after they scrapped the Titan core and accompanied APM83290/Gemini processor (which also sampled and had an evaluation board, btw).

    > sans AltiVec, pity

    Indeed. That's why I'm looking forward to news from Freescale. They still have 2.5 weeks left ;-)

    https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7001&forum=3&post_id=76983#76983

    > could a development board already exist?

    I very much guess so. From the press release linked before:

    "The company's Serengeti evaluation platform incorporates a dual core APM86290 running up to 1.5GHz and exposes all interfaces available on the SoC."
  • »14.12.10 - 10:07
    Profile
  • Paladin of the Pegasos
    Paladin of the Pegasos
    jcmarcos
    Posts: 1178 from 2003/3/13
    From: Pinto, Madrid ...
    Great! Let's make our own computer then!
  • »14.12.10 - 13:29
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > Let's make our own computer then!

    You mean using the eval board as is? While technically possible I'm afraid that could easily result in a system as expensive as the X1000 ;-)
  • »14.12.10 - 15:04
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    Ruud
    Posts: 335 from 2009/2/2
    From: Hampshire, UK
    As requested by Andreas_Wolf in another thread: dnetc -bench results for my Powermac G4 Quicksilver 2002 with 7448 @ 2GHz installed.

    dnetc v2.9105-511-CTR-09050816 for MorphOS (MorphOS 2.7).
    Please provide the *entire* version descriptor when submitting bug reports.
    The distributed.net bug report pages are at http://bugs.distributed.net/

    [Dec 21 18:27:45 UTC] Automatic processor type detection found
    a PowerPC 7448 (G4) processor.
    [Dec 21 18:27:45 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar).
    [Dec 21 18:28:04 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar)
    0.00:00:16.71 [19,916,193 nodes/sec]
    [Dec 21 18:28:04 UTC] OGR-NG: using core #1 (KOGE 3.1 Hybrid).
    [Dec 21 18:28:23 UTC] OGR-NG: Benchmark for core #1 (KOGE 3.1 Hybrid)
    0.00:00:16.31 [41,269,521 nodes/sec]
    [Dec 21 18:28:23 UTC] OGR-NG benchmark summary :
    Default core : #1 (KOGE 3.1 Hybrid)
    Fastest core : #1 (KOGE 3.1 Hybrid)
    [Dec 21 18:28:23 UTC] RC5-72: using core #0 (MH 2-pipe).
    [Dec 21 18:28:43 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe)
    0.00:00:17.00 [7,746,782 keys/sec]
    [Dec 21 18:28:43 UTC] RC5-72: using core #1 (KKS 2-pipe).
    [Dec 21 18:29:02 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe)
    0.00:00:16.08 [7,643,136 keys/sec]
    [Dec 21 18:29:02 UTC] RC5-72: using core #2 (KKS 604e).
    [Dec 21 18:29:21 UTC] RC5-72: Benchmark for core #2 (KKS 604e)
    0.00:00:16.94 [7,790,965 keys/sec]
    [Dec 21 18:29:21 UTC] RC5-72: using core #3 (KKS 7400).
    [Dec 21 18:29:40 UTC] RC5-72: Benchmark for core #3 (KKS 7400)
    0.00:00:16.11 [18,096,530 keys/sec]
    [Dec 21 18:29:40 UTC] RC5-72: using core #4 (KKS 7450).
    [Dec 21 18:29:59 UTC] RC5-72: Benchmark for core #4 (KKS 7450)
    0.00:00:16.67 [21,062,902 keys/sec]
    [Dec 21 18:29:59 UTC] RC5-72: using core #5 (MH 1-pipe).
    [Dec 21 18:30:17 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe)
    0.00:00:16.60 [7,074,547 keys/sec]
    [Dec 21 18:30:17 UTC] RC5-72: using core #6 (MH 1-pipe 604e).
    [Dec 21 18:30:36 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e)
    0.00:00:16.78 [6,999,792 keys/sec]
    [Dec 21 18:30:36 UTC] RC5-72 benchmark summary :
    Default core : #4 (KKS 7450)
    Fastest core : #4 (KKS 7450)

    [ Edited by Ruud on 2010/12/21 18:36 ]

    [ Edited by Ruud on 2010/12/21 18:38 ]

    [ Edited by Ruud on 2010/12/21 19:08 ]
    "We live, we die, we laugh, we cry"
  • »21.12.10 - 17:34
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    > As requested by Andreas_Wolf in another thread: dnetc -bench results for my
    > Powermac G4 Quicksilver 2002 with 7448 @ 2GHz installed.

    Thanks. Just in case you wonder: the purpose of asking you for these benchmark results is that I want to compare the 7448 to the 7447A/B on a clock per clock basis to see how much faster the 7448 is in dnetc, if at all.

    > dnetc v2.9105-511-CTR-09050816 for MorphOS

    That's not the most recent version for MorphOS (I used v2.9106-514) but I think it doesn't matter.

    Comparison:

    KOGE 3.1 Scalar: 7448 is 0.4% faster.
    KOGE 3.1 Hybrid: 7447A/B is 1.0% faster.
    MH 2-pipe: 7448 is 0.8% faster.
    KKS 2-pipe: 7448 is 1.5% faster.
    KKS 604e: 7448 is 1.2% faster.
    KKS 7400: 7448 is 1.1% faster
    KKS 7450: 7448 is 0.6% faster.
    MH 1-pipe: 7448 is 0.8% faster.
    MH 1-pipe 604e: 7448 is 1.0% faster.

    All in all I conclude that regarding dnetc there's no significant clock per clock difference between 7447A/B and 7448.
  • »21.12.10 - 20:13
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 12025 from 2003/5/22
    From: Germany
    Addendum:

    > "AltiVec SIMD has long been recognized for its performance in the
    > PowerQUICC processor line"
    > http://www.freescale.com/webapp/sps/site/homepage.jsp?code=QORIQ_HOME
    >
    > AltiVec in PowerQUICC? Huh? While it's true that they present QorIQ P1 to P3
    > as PowerQUICC replacements and QorIQ P4 and P5 as MPC86xx replacements
    > that surely doesn't make MPC86xx a PowerQUICC, does it? Are they now trying
    > to retroactively subsume MPC86xx (or even MPC74xx) as PowerQUICC?

    Just discovered that Freescale indeed lists the MPC8610 as PowerQUICC III (while MPC864x is listed as "Host Processor") there:

    http://www.freescale.com/webapp/sps/site/overview.jsp?code=PRDCT_LONGEVITY_HM

    Furthermore, MPC74xx and MPC86xx are listed in Freescale's 'PowerQUICC and QorIQ Processor Selector Guide':

    http://www.freescale.com/files/32bit/doc/brochure/PWRARCHQIQSG.pdf

    Explanation that adds up, anybody?
  • »29.12.10 - 19:52
    Profile