All times are UTC-06:00




Post new topic  Reply to topic  [ 49 posts ] 
Author Message
 Post subject:
PostPosted: Fri Oct 19, 2007 3:15 am 
Offline

Joined: Tue Nov 02, 2004 2:11 am
Posts: 161
Hallo Arno,
Quote:
>I have ars testbench 2.0 data now for MPC8641D, at 1.8 Ghz.
>On almost everything it is outrunning the dual (not dualcore) 1.8 970.

According to this paper the 1.8 GHz version was a shaming 70 W monster.
Motorola's original target for a dualcore G4 was the 15-25 W typical power range.
That's why they don't provide a 1.8 GHz version.

Lets summarize the above:
The Freescale 86xx seems to be performance wise a very capably CPU. It plays in the same performance league as the 970 or X86-Athlon/Opteron.
- The chip can be clocked with 1.8 Ghz (maybe more?)
- The early versions did consume 70 Watt at 1.8 Ghz

While 70 watt is a lot for embedded usage and too much for Genesi's cool computing ideas -
a power usage of 70 watt is normal for a desktop CPU. Most x86 system consume more.

And maybe Freescale can reach their original very low power consumption goal with the next generation of the CPU.



Quote:
>>The blade has two complete memory busses each with interleaved DIMS. So you need always to have 4 DIMS to be equipped in this board.

>To be precise, it has 128 Bit memory access, so a memory feature consists of (just) two DIMMs.

>>If I understand IBM's test on the JS21 blade correctly, then they have achieved this very good result using a board with two 970FX CPU chips

>970MP (only one core alive)

>>each with its own hypertransport ram bus. So the results of 7GB/sec was for both busses together.

>Yes, only it's not Hypertransport but IBM's Elastic Interfaces with plenty of bandwidth to the CPC945 northbridge
while Freescale limits throughput by using a single MPX bus for two G4 cores (with a single load/store unit each).
So you could say the IBM solution is RAM bandwidth limited while Freescale has a FSB limitation.
Supposedly the Power.org G5 devmachine can use DDR2-667, up from JS21's DDR2-533.
I think we are saying the same, aren't we?

The 970 is a very powerful CPU.
The 7 GB/sec is a very got result!
this good result was achieved with a system using two 970 CPU chips each with two memory Dimms = Two CPU chips + 4 DIMS total.

From what I can see a single chip DUAL 970 has the same memory through as a single chip 8641D ?
Quote:
>>Sergei's test showed a result 1.5 MB per CPU core with 400 MHz memory.

>And just over 2 GB/s for both cores - they don't scale 100% >for reasons mentioned above.
Sergei's test showed a result of 3 GB/s for both cores!
Please have a look at (page 16) again.
And please mind that the scale in Sergeis test is logarithmic.

I think the 3 GB/sec is not bad - its in the same range as the IBM CPU. The 3 GB/sec was with 400 MHz memory. I assume that if you use 600 Mhz memory the limit for both cores will be close to 4 GB/sec.
Quote:
>According to your site an iBook with 142 MHZ FSB exhibits 958 MB/s already...
The Ibook has nearly 1 Gb/sec with DDR 288 memory.

Yes, and the 86xx seems to be 3 times faster using DDR 400 memory.
I think it will be 4 times faster with DDR 600 memory.

I think the 86xx is a nice improvement over the G4.
Quote:
>>And IBM has the trio of super chips with: Power7
>Isn't Power7 a 2010 GA chip? :-)
Sorry, I meant Power6.


Cheers
Gunnar


Top
   
 Post subject:
PostPosted: Fri Oct 19, 2007 6:23 am 
Offline

Joined: Tue Nov 02, 2004 2:11 am
Posts: 161
Quote:
We will ship Gunnar one of our 8641D based systems. He has some interesting plans for it and we wish him the best on his Project.

R&B :)
Raquell and Bill,

Many thanks !!!


Top
   
 Post subject:
PostPosted: Fri Oct 19, 2007 1:57 pm 
Offline

Joined: Fri Sep 24, 2004 1:39 am
Posts: 111
>this good result was achieved with a system using two 970 CPU chips
>each with two memory Dimms = Two CPU chips + 4 DIMS total

It's not really the CPU that "owns" any DIMMs (like NUMA systems) - afaik they are shared and connected to the northbridge in 128 Bit access configuration (combines two 64 Bit RAM channels).

Each CPU chip has an Elastic Interface to the northbridge. This means 32 Bit unidirectional up- and downstreams with half the CPU clock datarate:

2700/2 Megatransfers/s * 4 Byte = 5400 MB/s each direction per chip, minus the address/control traffic overhead.

>Sergei's test showed a result of 3 GB/s for both cores!
>Please have a look at (page 16) again.

Alright, you were referring to a calculated limit while I had a look at page 20.

Gunnar, let's stop the discussion here, we are looking forward to seeing your results with the Genesi-provided 8641D board! :-)


Top
   
 Post subject:
PostPosted: Sun Oct 21, 2007 7:26 am 
Offline
Genesi

Joined: Fri Sep 24, 2004 1:39 am
Posts: 1422
Yes, in the meanwhile, some of the original work is posted here:

http://www.freevec.org/benchmark.php

http://www.freevec.org/features.php

Konstantinos and Gunnar worked on some of this together.

R&B :)

_________________
http://bbrv.blogspot.com


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 49 posts ] 

All times are UTC-06:00


Who is online

Users browsing this forum: No registered users and 6 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
PowerDeveloper.org: Copyright © 2004-2012, Genesi USA, Inc. The Power Architecture and Power.org wordmarks and the Power and Power.org logos and related marks are trademarks and service marks licensed by Power.org.
All other names and trademarks used are property of their respective owners. Privacy Policy
Powered by phpBB® Forum Software © phpBB Group