Poster | Thread |
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 31-Oct-2013 9:50:22
| | [ #181 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @broadblues
Maybe there should be some effort from those who know to tutor those who don't, then _________________ 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. |
|
Status: Offline |
|
|
NutsAboutAmiga
| |
Re: Next Freescale high performance PPC chip. Posted on 31-Oct-2013 15:53:19
| | [ #182 ] |
|
|
|
Elite Member |
Joined: 9-Jun-2004 Posts: 12795
From: Norway | | |
|
| |
Status: Online! |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 18-Feb-2014 7:19:00
| | [ #183 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| @thread
New convergence chip: http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=B3421 (not suitable for our niche, btw. but the first device with e6501 cores)
Perhaps some T3 chip will be announced next...
a snippet from there somewhere: "10 million units of QorIQ processing platforms shipped with 140% growth year over year (2013 vs. 2012) in the multicore segment" Good for future R&D. Last edited by KimmoK on 18-Feb-2014 at 07:34 AM. Last edited by KimmoK on 18-Feb-2014 at 07:29 AM.
_________________ - KimmoK // For freedom, for honor, for AMIGA // // Thing that I should find more time for: CC64 - 64bit Community Computer? |
|
Status: Offline |
|
|
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 18-Feb-2014 9:47:01
| | [ #184 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @KimmoK
Not sure why you feel it absolutely couldn't be used for a generic desktop board, with 2x PCIe 2.0, 2x SATA2, USB and ethernet.
Two dual-thread cores with Altivec is exactly the thing I wanted. The DSP cores should also be exploitable for some good fun, like video codecs etc. Or use them as copper/blitter or something. Imagination is the limit here. _________________ 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. |
|
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 18-Feb-2014 10:19:05
| | [ #185 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| @olegil
Ok, it can be used. (but most likely there are better/cheaper options)
UPDATE: btw. T1040 (and T1020) reference boards should come out around April for "the rest of the world". Mass production of T1040 should start Q4 or so. Next month everyone can get samples (currently only for bigger/better customers), etc, etc. (need to get my NDA sorted out ...) Last edited by KimmoK on 18-Feb-2014 at 01:00 PM.
_________________ - KimmoK // For freedom, for honor, for AMIGA // // Thing that I should find more time for: CC64 - 64bit Community Computer? |
|
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 8-Apr-2014 15:30:42
| | [ #186 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| |
Status: Offline |
|
|
AmigaBlitter
| |
Re: Next Freescale high performance PPC chip. Posted on 8-Apr-2014 16:26:50
| | [ #187 ] |
|
|
|
Elite Member |
Joined: 26-Sep-2005 Posts: 3512
From: Unknown | | |
|
| @KimmoK
interesting for an upgradable system:
"The T4240, with 24 virtual cores, is the flagship of the QorIQ T series. Joined by the T4160 (16 virtual cores) and T4080 (eight virtual cores), the T4 family has a 3x performance scaling factor within a pin compatible package"
_________________ retired |
|
Status: Offline |
|
|
billt
| |
Re: Next Freescale high performance PPC chip. Posted on 8-Apr-2014 20:48:20
| | [ #188 ] |
|
|
|
Elite Member |
Joined: 24-Oct-2003 Posts: 3205
From: Maryland, USA | | |
|
| @AmigaBlitter
I'm not sure how well the pin compatibility makes for an upgradeable system. This doesn't go into a socket, remember, this is a BGA. That's not easy to change, and has a very small number of possible changes before the PCB fails from the stresses of BGA rework process. It's hard to guarantee that the first time will be a success.
In order for such things to really be upgradeable, I'd really encourage one to consider board modularity, such as COM-Express, which is quite similar in concept to what many accelerator cards really were in classic Amigas. _________________ All glory to the Hypnotoad! |
|
Status: Offline |
|
|
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 0:00:03
| | [ #189 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @billt
The T4080 was announced recently?
More to the point, bga sockets do exist. Economic viability uncertain, though. _________________ 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. |
|
Status: Offline |
|
|
billt
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 4:17:55
| | [ #190 ] |
|
|
|
Elite Member |
Joined: 24-Oct-2003 Posts: 3205
From: Maryland, USA | | |
|
| |
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 6:14:48
| | [ #191 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| T1023 & 24 stuff: http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=T1024
Hmmm... spotted also T1014 ... more info in the pipeline, perhaps.
(got Aava Mobile job secured for next three moths, and now I do not have training + job in parallel, so I hope I now can have a little more time for other stuff... too bad the pay is not exactly what I wanted ... need to save for T10xxRDB for a while...)
btw... audio might be connectable via this: "QUICC Engine TDM". (At least SSI is done via TDM on some P series chips, unless I'm mistaken.)
Last edited by KimmoK on 09-Apr-2014 at 06:27 AM. Last edited by KimmoK on 09-Apr-2014 at 06:17 AM. Last edited by KimmoK on 09-Apr-2014 at 06:15 AM.
_________________ - KimmoK // For freedom, for honor, for AMIGA // // Thing that I should find more time for: CC64 - 64bit Community Computer? |
|
Status: Offline |
|
|
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 7:40:53
| | [ #192 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @KimmoK
SSI? You mean I2S? Quick Engine doesn't support I2S, but it MIGHT be possible to turn the QE SPI (which supports back-to-back data transfers) into a DSP-compatible audio stream using some external logic. Hope I might some day fairly soon be able to say a definite yes or no to this.
There's also a rumour that a company somewhere can be hired to write things like I2S support for QE, given enough money. If the answer to the first question above is "no" then I'll be sure to investigate this as well. _________________ 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. |
|
Status: Offline |
|
|
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 7:46:50
| | [ #193 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @billt
Ah, so they announced it yesterday. Thought I had somehow missed this while taking care of mother and newborn a couple of weeks back.
The T4080 seems a pretty good match for an Amiga.
My biggest problem with COM Express is that it's made with a different architecture in mind, so it doesn't fit THAT well with a T4xx0.
I would think that if the motherboard is made cheap enough (as in, it's a SoC and the clocks, logic and PSU parts needed to run a bunch of slots) then upgrading the CPU becomes a rather moot point. However, if you start including all sorts of cost-raising stuff on there, then of course people are going to want to have an option of keeping the board if they fry their CPU. _________________ 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. |
|
Status: Offline |
|
|
billt
| |
Re: Next Freescale high performance PPC chip. Posted on 9-Apr-2014 15:18:13
| | [ #194 ] |
|
|
|
Elite Member |
Joined: 24-Oct-2003 Posts: 3205
From: Maryland, USA | | |
|
| @olegil
Yes, COM-Express was designed essentially to provide what's expected in any garden variety x86 PC. I don't think that's necessarily a problem. There are already several PPC modules for COM-Express, though some, particularly earlier ones, have taken some artistic liberty with the Kontron/PICMG pinout definitions. But even some Freescale eval systems are COM-Express now, and Freescale and someone else have teamed up to work on a new type definition that better fits an embedded sort of system, and are petitioning for official recognition of this in a future PICMG release of COM.0 Standard.
Expanding the COM Express® Specificationfor Alternative Microprocessor Architectures
Freescale COM-Express Carrier schematics
and Carrier Errata
and Freescale COM-Express"Fact Sheet"
My own list of PowerPC on COM-Express modules (some staleness may apply by now)
Yea, I've been obsessed with this format for 10 years now...
And I was disappointed that the T4240 didn't have a COM-Express based eval board. :/ I would have liked one of those.
Fitting a T4240 family part, or other current PPC SoC, onto this, there would be things that might not get a pinout. I don't want a Router for my Amiga, so I am not really concerned that so many ethernet ports on the CPU chip are no-connects. For what we Amigans would use, I think that a T4* paired with a southbridge like found in the X1000, should make good use of the COM-Express Type 6, or maybe Type 2 if you prefer, which are the two most normal x86 PC definitions, to make for a modular "Amiga" system. Last edited by billt on 09-Apr-2014 at 03:28 PM. Last edited by billt on 09-Apr-2014 at 03:28 PM. Last edited by billt on 09-Apr-2014 at 03:19 PM.
_________________ All glory to the Hypnotoad! |
|
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 28-Apr-2014 7:52:10
| | [ #195 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| |
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 29-Apr-2014 10:28:05
| | [ #196 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| Some prices are out: T1042 USD65...USD80 T2081 USD130 T2080 USD142
I imagine T1022 will be very attractively priced...
Those who sit on top of their PA6T stock should take a notice, they can not ask USD1000+ per chip any more.
UPDATE, reference boards: T1042RDB-PA $1,098 T2080RDB-PA $1,620 T4240RDB-PB $1,562
hmmm... what's my excuse now. (will get price for myself next week I believe)
btw. next gen freescale chips seem to make AMCC&APM chips totally obsolete... I wonder if ACube produces many PPC460 based boards any more...
Update: another video ... http://www.youtube.com/watch?v=_W53LzpUrIM T1 & T2 Last edited by KimmoK on 29-Apr-2014 at 11:57 AM. Last edited by KimmoK on 29-Apr-2014 at 10:43 AM. Last edited by KimmoK on 29-Apr-2014 at 10:36 AM. Last edited by KimmoK on 29-Apr-2014 at 10:31 AM.
_________________ - KimmoK // For freedom, for honor, for AMIGA // // Thing that I should find more time for: CC64 - 64bit Community Computer? |
|
Status: Offline |
|
|
olegil
| |
Re: Next Freescale high performance PPC chip. Posted on 29-Apr-2014 12:24:54
| | [ #197 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @KimmoK
I was about to say "nice", but "nice enough" should cover most bases, I think.
A sales representative of some company once asked me the stupidest question ever: "what's your target price?", to which I obviously replied "well, I would take it for free"
But in the real world, USD 130 for the T2081 is not at all bad, with it's quad SMT + Altivec cores. Unfortunately, how stupid are they, making a "pin compatible" chip that doesn't have the same feature set? It's the P3041/P4080 all over again. Morons.
So I would not even bother with that option, as it needs all sorts of high speed muxes and whatnots. Also, one family is DDR3/3L while the other is DDR3L/4, so that severely limits the options if you need to support both.
Now, going with DDR3L is a good plan anyway, since it means more reuse if you want to make two designs.
Plausible project plan: Step 1: Read both data sheets and reference manuals for BOTH families. Step 2: Read everything again. Step 3: Find common ground between T2080 and T10x2, do the floorplan based on this.Try to avoid depending greatly on features found in only one of them, as drivers won't get the focus they need. Step 4: Then pop in the T2080 in the design, connect it up and complete the design. Anything that can't fit has to be moved until it fits. Step 5: Take the T2080 out, replace with the T10x2. Now, any changes not related to the T10x2, it's power planes and it's decoupling capacitors, should be also backported to the T2080 version. Step 6: Send the T10x2 version into prototyping first. If you did a good enough job in step 2, this'll work. If not, there's gonna be a loop here until it does (or you give up) Step 7: Now incorporate any bugfixes you found in step 6 on the T2080 version, and send that one off to prototyping. Step 8: Profit (whew, I thought I would never get here). _________________ 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. |
|
Status: Offline |
|
|
AmigaBlitter
| |
Re: Next Freescale high performance PPC chip. Posted on 29-Apr-2014 13:07:52
| | [ #198 ] |
|
|
|
Elite Member |
Joined: 26-Sep-2005 Posts: 3512
From: Unknown | | |
|
| @olegil
I this "low end" chip are cheaper, maybe a low cost motherboards can be built around those chip, in order to have low, mid, and high end system.
Don't know why these chip have been not considered.
More important is that all the product line shares the same features, such altivec. This can avoid software version compilation problems (see mplayer) _________________ retired |
|
Status: Offline |
|
|
KimmoK
| |
Re: Next Freescale high performance PPC chip. Posted on 29-Apr-2014 13:33:36
| | [ #199 ] |
|
|
|
Elite Member |
Joined: 14-Mar-2003 Posts: 5211
From: Ylikiiminki, Finland | | |
|
| @AmigaBlitter
>Don't know why these chip have been not considered.
Because they only now are available.
>More important is that all the product line shares the same features, such altivec. This can avoid software version compilation problems (see mplayer)
Cyrus+ will not have altivec. T1022/1042 does not have altivec, but the same core as Cyrus+ (x5000/20).
I'm afraid some CPU differences must be taken into account in SW (same applies in x86 etc...).
Observations: T2080/2081 should be slightly cheaper than P3041 and up to half of the price of P5020 ... Looking at only P5020 vs PA6T prices (etc) ... Cyrus+ X5000/20 HW cost should be at least 700usd lower than x1000. ... And rather than trying to save some 150usd with P3041 and supporting e500mc core, perhaps they should rather port AOS4 to e6500 core & skip e500 series. (supporting e5500 is anyway a good thing, IMO) Last edited by KimmoK on 06-May-2014 at 07:41 AM. Last edited by KimmoK on 06-May-2014 at 07:38 AM.
_________________ - KimmoK // For freedom, for honor, for AMIGA // // Thing that I should find more time for: CC64 - 64bit Community Computer? |
|
Status: Offline |
|
|
tlosm
| |
Re: Next Freescale high performance PPC chip. Posted on 29-Apr-2014 16:43:00
| | [ #200 ] |
|
|
|
Elite Member |
Joined: 28-Jul-2012 Posts: 2746
From: Amiga land | | |
|
| @KimmoK
I see on freescale the 5021 have the Multimedia Acceleration (altivec?) probably there is the opportunity to have this on the X5000 when it is done
_________________ I love Amiga and new hope by AmigaNG A 500 + ; CDTV; CD32; PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB; MacBook Pro Retina I7 2.3ghz; #nomorea-eoninmyhome |
|
Status: Offline |
|
|