28-Apr-2024 07:38 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 94 items in your selection (but only 44 shown due to limitation) [1 - 50] [51 - 94]
[News] Marvell announces Discovery III NorthbridgeANN.lu
Posted on 30-Sep-2003 15:01 GMT by takemehomegrandma94 comments
View flat
View list
Marvell today announced the Discovery III northbridge. It features a 200MHz PowerPC CPU interface, 200MHz DDR SDRAM Interface (400Mbps data rate), Dual CPU SMP Support (MPX and 60x modes), PCI-X, Gigabit ethernet and is software compatible with other Discovery northbridges.

Source: morphos-news.de

Marvell announces Discovery III Northbridge : Comment 51 of 94ANN.lu
Posted by Anonymous on 01-Oct-2003 02:36 GMT
In reply to Comment 26 (Hagge):
>i think some g4s got ddr support Sure they do, the point is you won't get far with 4MB max...(hint: L3/private memory).
Marvell announces Discovery III Northbridge : Comment 52 of 94ANN.lu
Posted by Some Farker on 01-Oct-2003 03:13 GMT
"Personally, I hold by the Clarke - Sturgeon law:
90% of any sufficently advanced technology is
indistinguishable from crap."
Marvell announces Discovery III Northbridge : Comment 53 of 94ANN.lu
Posted by Andreas Wolf on 01-Oct-2003 03:51 GMT
In reply to Comment 29 (Joe "Floid" Kanowitz):
> looks like http://www.tundra.com/ is another company to keep an eye on.

I have been keeping an eye on them for some time and I wonder what happened to their Tsi890 northbridge:

http://www.tundra.com/page.cfm?TREE_ID=101036
http://www.tundra.com/page.cfm?TREE_ID=101049
http://web.archive.org/web/20011215151717/http://www.tundra.com/page.cfm?TREE_ID=101039

> Those Motorola designs haven't evolved since.. 1995?

AFAIK the MPC/Tsi106 is from 1996 and the MPC/Tsi107 is from 1999.
Marvell announces Discovery III Northbridge : Comment 54 of 94ANN.lu
Posted by Andreas Wolf on 01-Oct-2003 04:09 GMT
In reply to Comment 32 (Joe "Floid" Kanowitz):
> PCI Express announcements from graphics vendors. I think NVidia and ATI have
> both promised to go that road.

Yes, you're right.

http://www.nvidia.com/object/IO_20020416_7440.html
http://www.nvidia.com/object/IO_20020909_4673.html
http://www.nvidia.com/object/IO_8379.html
http://www.nvidia.com/object/pci_express.html

http://www.ati.com/companyinfo/press/2001/4405.html
http://www.ati.com/companyinfo/press/2002/4481.html
http://www.ati.com/companyinfo/press/2002/4532.html
http://www.ati.com/companyinfo/press/2003/4684.html

and also

http://www.3dlabs.com/whatsnew/pressreleases/pr03/03-02-18-pci.htm
Marvell announces Discovery III Northbridge : Comment 55 of 94ANN.lu
Posted by Anony Mouse on 01-Oct-2003 05:34 GMT
In reply to Comment 11 (Hagge):
> 16x PCI-X will replace AGP anyway before whatever replaces that later on ;)
No. PCI-X was never thought to replace AGP. I even don't know if a PCI-X graphic card is available on the market.

In the x86 world AGP should replaced by PCI Express (This is not PCI-X). First x86 Mainboards for PCI Express will come out in 2004.
Marvell announces Discovery III Northbridge : Comment 56 of 94ANN.lu
Posted by Anonymous on 01-Oct-2003 05:39 GMT
>Posted by Some Farker (67.8.237.55) on 01-Oct-2003 05:13:36>
>"Personally, I hold by the Clarke - Sturgeon law:
>90% of any sufficently advanced technology is
>indistinguishable from crap."

I like that attitude.
This law should be enhanced by adding two other percentages:
-the count/percentage of people that are incapable of handling "modern" technology
-the count/percentage of people that earn a living on helping above people
at handling "modern" technology
Marvell announces Discovery III Northbridge : Comment 57 of 94ANN.lu
Posted by Anonymous on 01-Oct-2003 05:48 GMT
Besides, the original Clarke law states:

"Any sufficiently advanced technology is indistinguishable from magic."
Thats also a good attitude.

So anything that gives a big hassle cant be advanced. We knew that since
early Windows (3.x) days, didn't we.

Send...oops:

"Using Abusive language and swear words, or insulting other persons is not tolerated, and any such comments can be deleted with any further warning. You are going to add a comment to a site which is also being viewed by minors or people who are easily offended. Please act accordingly"

So my use of "Windows" was an offense? Ok, I agree.
Marvell announces Discovery III Northbridge : Comment 58 of 94ANN.lu
Posted by Crumb // AAT on 01-Oct-2003 05:48 GMT
In reply to Comment 33 (Eva):
"-----
the sad part is that g3/g4s still don't have DDR support
-----
???? Also Discovery 2 supports perfectly DDR ram! Read well the specs! "200MHz DDR SDRAM Interface (400Mbps data rate)"
It uses the same Aplle G5 Bus approach!
... I have to admint that for my bros it's a Apple G5 BUS copy!
Anyway really the best open chipset for PPC systems
And Bplan already use a chip Discovery for Pegasos 2 and with the backward compatibility of Discovery 3 chipsets ... :D"

@Eva:
Even if the northbridge supported 1Ghz DDR memory, the g3/g4 wouldn't take advantage of it because they don't support DDR. They would continue talking with the northbridge in the same way the would do with an older northbridge.

note that it's great to be able to use 400Mhz DDR memory because the rest of the board can take advantage of it, but sorry, until IBM or Motorola decides to add DDR support g3/g4 will run as good as with normal sdram. Simply check out benchmarks of macs with g4 and DDR memory and G4 with normal SDRAM... you will see that there's little speed increase.

There were rumours of some very limited test editions of G4s cpus with DDR support and apple prototypes, but they are probably false... :-/
Marvell announces Discovery III Northbridge : Comment 59 of 94ANN.lu
Posted by Anonymous on 01-Oct-2003 07:05 GMT
In reply to Comment 1 (Bob):
Don't you just love it when someone posts without knowing what they are on about, Bob?

Had you stopped to think for a second, you would have realised that it already does support the DDR400 which you demand - so can we look forward to you withdrawing your condemnation of it?
Marvell announces Discovery III Northbridge : Comment 60 of 94ANN.lu
Posted by Eva on 01-Oct-2003 08:24 GMT
In reply to Comment 30 (Miffo):
---------
Posted by Miffo (193.11.48.71) on 30-Sep-2003 20:19:51
In Reply to Comment 20:
Cool!

But it is not as small as the Amiga One Lite!
;-)
----------
Ehm ... this is a DUAL CPU board ...
Marvell announces Discovery III Northbridge : Comment 61 of 94ANN.lu
Posted by Eva on 01-Oct-2003 08:35 GMT
In reply to Comment 58 (Crumb // AAT):
------
until IBM or Motorola decides to add DDR support g3/g4 will run as good as with normal sdram.
------
Oky I just understand your point of you ... better Mai chip in AMigaOne (with all the difficulties to find Sdram that works and cost less than DDR one) :D

Anyway if you see the benchmark I can tell you also that ... if you compare an ATi 8500 PCI with an ATi 8500 AGP, the differences in speed are little (in Q3 tests about 6-7 fps) ... this means that Agp is useless?
I think no.
Point out that "DIscovery DDR support is useless" is simply not true.
Both the machine speed and the user benefits are rewarded by DDR support.
In these dayz Singapore dealers Sdram price are higher than DDR 333 one! And it's also difficult to find Sdram around. So also the simple fact that we can INSTALL DDR-ram is a step foward.
Marvell announces Discovery III Northbridge : Comment 62 of 94ANN.lu
Posted by minator on 01-Oct-2003 09:07 GMT
In reply to Comment 61 (Eva):
Eva wrote:

>Point out that "DIscovery DDR support is useless" is simply not true.
>Both the machine speed and the user benefits are rewarded by DDR support.
>In these dayz Singapore dealers Sdram price are higher than DDR 333 one! And it's
>also difficult to find Sdram around. So also the simple fact that we can INSTALL
>DDR-ram is a step foward.

It more than that, I fully expect memory speeds to go up.

The DDR bus means the read from memory takes less time and can get buffered in the core of the North Bridge. So, if the CPU is reading a big chunk from memory the second and subsequent read should come from the buffer and this will reduce the access latency.

Memory access is quite a bit below it's theoretical values but this will give the CPU a nice boost in memory speed even if it is using the same speed CPU-NorthBridge bus.

This also applies to PCs, think your PC 2700 memory is actually reading at 2.7GBytes a second, think again, it's not even reading at the theoretical maximum of PC133 SDRAM.
Marvell announces Discovery III Northbridge : Comment 63 of 94ANN.lu
Posted by Crumb // AAT on 01-Oct-2003 09:51 GMT
In reply to Comment 60 (Eva):
you can fit a dual megarray module in an amiga One Lite, although I'm not sure if it would fit in all cases... neither MorphOS or OS4 can take advantage of a multicpu setup, and I'm not interested in Linux (it sucks) so, I can't see the lack of multicpu as a big problem... moreover, I've used a single cpu pegasos and I'm quite happy with it... one cpu is enough for me by the moment...
Marvell announces Discovery III Northbridge : Comment 64 of 94ANN.lu
Posted by Crumb // AAT on 01-Oct-2003 10:03 GMT
In reply to Comment 61 (Eva):
"Anyway if you see the benchmark I can tell you also that ... if you compare an ATi 8500 PCI with an ATi 8500 AGP, the differences in speed are little (in Q3 tests about 6-7 fps) ... this means that Agp is useless?"

no, I'm not saying DDR support is useless... now the prices with normal sdram are matched but in a few quarters the prices of normal sdram will rise and and ddr ones will fall.

Regarding the AGP vs PCI... For me the AGP was only developed to sell new graphic cards as they did changing cpu sockets (firstly slot1 was great, a little later it wasn't), etc... I would have prefered an improvement of the whole pci bus (a release of pci3, etc...). I agree that the possibility of accessing system ram is great (but it can be achieved with a bus master pci card, can't it?), but all that bandwitch won't help much... I prefer a good graphic core with lots of ram in pci than a super agp x8 board with a crappy core...

Even if the Pegasos II was released with only PCI I wouldn't mind much... it's still a kick-ass server with its 3 ethernet gigabits, you have agp<--->pci adapters available, in AmigaOS/MorphOS you don't really take advantage of modern gfx cards, so even a radeon 7500 pci will be enough...

"Point out that "DIscovery DDR support is useless" is simply not true. "

I don't remember saying that. I'm only say that it's a pity that the FSB of the cpu doesn't support DDR transfers so the memory's extra bandwitch is used...

"In these dayz Singapore dealers Sdram price are higher than DDR 333 one! And it's also difficult to find Sdram around. So also the simple fact that we can INSTALL DDR-ram is a step foward."

In spain sdram is still cheaper in "Alternate" and other shops, but we can be sure that in a few months the prices will change a lot.

"So also the simple fact that we can INSTALL DDR-ram is a step foward."

I haven't said the contrary, but it would be a greater step if motorola or ibm decided to add DDR support to its range of g3/G4s...
Marvell announces Discovery III Northbridge : Comment 65 of 94ANN.lu
Posted by Mirror on 01-Oct-2003 11:30 GMT
In reply to Comment 20 (takemehomegrandma):
[quote]
But it's there allright!

http://www.bplan-gmbh.de/gfx/pegasos2.jpg

(hehe ;-) )
[/quote]

I hope that is not a recent photographs. Notice how the bottoms have of the picture has been conviently cropped to hide all the wire fixes running towards the bottom of the board

Is that April 3 ?
Marvell announces Discovery III Northbridge : Comment 66 of 94ANN.lu
Posted by Anonymous on 01-Oct-2003 11:42 GMT
In reply to Comment 52 (Some Farker):
when you quote something, quote it right

replace crap with magic
Marvell announces Discovery III Northbridge : Comment 67 of 94ANN.lu
Posted by tarbos on 01-Oct-2003 11:44 GMT
In reply to Comment 64 (Crumb // AAT):
>Even if the Pegasos II was released with only PCI I wouldn't mind much...>it's still a kick-ass server with its 3 ethernet gigabits With one gigabit port you mean?
Marvell announces Discovery III Northbridge : Comment 68 of 94ANN.lu
Posted by Joe "Floid" Kanowitz on 01-Oct-2003 12:04 GMT
In reply to Comment 64 (Crumb // AAT):
> I haven't said the contrary, but it would be a greater step if motorola or
> ibm decided to add DDR support to its range of g3/G4s...

Okay, come off it, guys. The *throughput* of the bus is what matters in selecting a memory technology, not the clocking. Half the point of *having* a northbridge (in conventional, non-integrated designs) is to tie the CPU to memory that, these days, is almost always going to be completely unrelated, be it as simple as an asynchronous clocking or as complex as a move to MRAM or the next big miracle technology.

To approach the theoretical maximum of a single-rate 200MHz bus, you'd need 200MHz single-rate SDRAM, which doesn't exist on the common market. The DDR nomenclature that revolves around clock rates (DDR333, DDR400...) is the mythical number - DDR333 runs at 166MHz. So you 'only' waste 133 mythical MHz, which probably help drive down latencies, keep the caches packed and prevent PCI/AGP from starving the CPU anyway.

Sure, it'd be great to have stonkingly faster CPU buses, and DDR and QDR have been the way to achieve them for now. But that's just one popular variety of 'magic,' and DDR and QDR on the CPU have nothing direct to do with DDR and QDR on the SDRAM.

Whine about the throughputs and latencies, please, not the pixie dust. If you set out to redesign the 60x/MPX buses for DDR (which they were supposed to have been doing with MPX+ on the never-unveiled? G4 MPC7470) and require a reworking of all northbridges anyway, you may as well ditch the old protocol and move to something more scalable overall. (Which they've done; 970 uses the ApplePI 'Elastic Bus,' allowing per-CPU links to the memory controller; the enlightened side of the x86 camp went NUMA and HyperTransport, and so on.)

That's harsh, I'm overcaffeinated. Again, it'd be wonderful if they made chips that supported perfect, future-proof FSBs, but the fact is they don't, and one assumes they would if it were as easy or economical as we all wish it were. We all know the answer, we just don't like to hear it -- the only way to help that situation is by growing the market and making it more worthwhile for the designers to overcome the hurdles. If the performance crown is all that matters, well... you can get three computers for the price of a computer, these days. Six if you count Hyperthreading.
Marvell announces Discovery III Northbridge : Comment 69 of 94ANN.lu
Posted by tonya on 01-Oct-2003 12:35 GMT
eva: and what good does this give us ? ..i mean dual cpu or not.., i cant afford something that will cost 4000 USD! , people who buy this over other alternatives atm would have to be rather rich , beyond that ..last i checked there is no OS to be runned with such an hw.

if there is an OS in progress for such, it will take YEARS for it to succeced , look at the x86 plattform.., smp died faster than anything else, mainly due to the os's who said we support it but we dont promise any actual speedup's .

but i agree some chipset designers pulled it off quite well but it didnt really make em for "normal users" .
'
Marvell announces Discovery III Northbridge : Comment 70 of 94ANN.lu
Posted by Andreas Wolf on 01-Oct-2003 12:39 GMT
In reply to Comment 62 (minator):
http://www.barefeats.com/pmddr.html and http://www.barefeats.com/pmddr2.html suggest otherwise.
Marvell announces Discovery III Northbridge : Comment 71 of 94ANN.lu
Posted by Andreas Wolf on 01-Oct-2003 12:52 GMT
In reply to Comment 65 (Mirror):
It was already explained in this thread what this picture shows. Hint: It's not the PegasosII.
Marvell announces Discovery III Northbridge : Comment 72 of 94ANN.lu
Posted by Alkis Tsapanidis on 01-Oct-2003 14:30 GMT
In reply to Comment 35 (Kulwant Bhogal):
Ehm, only NForce boards have Dual DDR.
Marvell announces Discovery III Northbridge : Comment 73 of 94ANN.lu
Posted by Alkis Tsapanidis on 01-Oct-2003 14:44 GMT
In reply to Comment 65 (Mirror):
That photo shows a Pegasos1 prototype. That was even before the Pegasos PCB
was printed, let alone April.
Marvell announces Discovery III Northbridge : Comment 74 of 94ANN.lu
Posted by Kolbjørn Barmen on 01-Oct-2003 15:00 GMT
In reply to Comment 67 (tarbos):
Heh, only one? I thought it was supposed to be three?

Oh well, that's was the end of the low profile pegasos router project I guess.

Hehe
Marvell announces Discovery III Northbridge : Comment 75 of 94ANN.lu
Posted by takemehomegrandma on 01-Oct-2003 18:11 GMT
In reply to Comment 30 (Miffo):
> Cool!

> But it is not as small as the Amiga One Lite!

Dear Newbie, I am sorry for that. That was in fact a "Pegasos 0"!
That was the reason for the "(hehe ;-) )" part at the bottom ...

;-)
Marvell announces Discovery III Northbridge : Comment 76 of 94ANN.lu
Posted by Andreas Wolf on 01-Oct-2003 19:13 GMT
In reply to Comment 74 (Kolbjørn Barmen):
When the PegasosII was first announced it was supposed to get three GBit ports. Shortly after they changed to two ports. And in August they mentioned only one port at German TV.
Marvell announces Discovery III Northbridge : Comment 77 of 94ANN.lu
Posted by Kulwant Bhogal on 01-Oct-2003 19:13 GMT
In reply to Comment 72 (Alkis Tsapanidis):
Er, what's your definition of Dual DDR?

To me it means Dual Channel DDR. I built a PC for my sister using a ASUS P4P8X - which uses the Intel 865 Chipset (clearly not nForce) and supports upto DDR400 in dual channel mode. There is also an Intel 875 chipset and I think SiS, VIA may also have Dual channel DDR capable chipsets for the P-IV (although I haven't checked).

KB
Marvell announces Discovery III Northbridge : Comment 78 of 94ANN.lu
Posted by takemehomegrandma on 01-Oct-2003 20:33 GMT
In reply to Comment 76 (Andreas Wolf):
@ Andreas Wolf

What do you think about this?

1 AGP Port (2x/4x/8x)
USB 2.0 Ports
3 PCI Sockets
Firewire (1EEE-1394a)
Ethernet (10/100baseT)
IrDA
S/PDIF Sound Jack
Standard Microphone, Sound In, and Sound Out jacks
MIDI/Game Port
Serial and Parallel ports
Marvell announces Discovery III Northbridge : Comment 79 of 94ANN.lu
Posted by hammer on 01-Oct-2003 21:45 GMT
In reply to Comment 61 (Eva):
>Anyway if you see the benchmark I can tell you also that ... if you compare >an ATi 8500 PCI with an ATi 8500 AGP, the differences in speed are little (in >Q3 tests about 6-7 fps) ... this means that Agp is useless?

IF one exceed video card's on-board memory AGP transfers can be important (e.g. UT2003 with custom map called DM-2019). The main memory can be use for extra AGP texture memory. Q3 test doesn’t really stress the power of the current high performance hardware.
Marvell announces Discovery III Northbridge : Comment 80 of 94ANN.lu
Posted by hammer on 01-Oct-2003 22:06 GMT
In reply to Comment 72 (Alkis Tsapanidis):
Note that the new flagship platforms such as the Athlon FX51 and Opteron have dual channel memory controllers. AMD64’s northbridge and CPU FSB runs at full speed with the CPU core e.g. AMD64 core @2.2Ghz = Northbridge @2.2 Ghz FSB. It has already crushed the competing 1Ghz FSB Northbridge<>CPU from another company.
Marvell announces Discovery III Northbridge : Comment 81 of 94ANN.lu
Posted by hammer on 01-Oct-2003 22:14 GMT
In reply to Comment 62 (minator):
>This also applies to PCs, think your PC 2700 memory is actually reading at >2.7GBytes a second, think again, it's not even reading at the theoretical >maximum of PC133 SDRAM.
SISoft Sandra 2003 Pro benchmark indicates otherwise...
Marvell announces Discovery III Northbridge : Comment 82 of 94ANN.lu
Posted by Andreas Wolf on 02-Oct-2003 00:55 GMT
In reply to Comment 78 (takemehomegrandma):
Where's the GBit Ethernet? :-)
Marvell announces Discovery III Northbridge : Comment 83 of 94ANN.lu
Posted by tarbos on 02-Oct-2003 01:19 GMT
In reply to Comment 74 (Kolbjørn Barmen):
>Oh well, that's was the end of the low profile pegasos router project I guess. How can a Slot1 design be low profile? :)
Marvell announces Discovery III Northbridge : Comment 84 of 94ANN.lu
Posted by tarbos on 02-Oct-2003 01:23 GMT
In reply to Comment 80 (hammer):
>AMD64’s northbridge And what may this be?
Marvell announces Discovery III Northbridge : Comment 85 of 94ANN.lu
Posted by Joe "Floid" Kanowitz on 02-Oct-2003 02:55 GMT
In reply to Comment 84 (tarbos):
tarbos said,
>>AMD64’s northbridge
>And what may this be?

I know (or hope) you're just playing, but he meant the on-die memory controller. For the even-less-technical sorts reading, it's a fair comparison; AMD64 does away with the northbridge as we know it, integrating half of it on the CPU itself... and the other half as well, if you count HyperTransport as the Uberbus it is. (The 'tunnels' or 'bridges' used to provide PCI, PCI-X, AGP, 3GIO, USB, legacy ports, IDE and so on are basically just HyperTransport peripherals. Think of a conventional CPU bus crossed with Rambus/Firewire/SCSI.)

The performance numbers speak for themselves; designing it that way certainly doesn't hurt. How much of that is really based on clock rate... I dunno. You need the clock rate in some respect to have the fat internal bandwidth that makes it all possible, but if my perspective is accurate, it's more that it helps remove the chance of an external chipset screwing things up (viz. Apple's discussion of MPX? reordering lower down). Or a bottlenecking bus, sure... but it's more a matter of *design* and parallelism than just clock-speed waving.

I'm really lost on roadmaps right now, but whatever ClawHammer was/is, it was once supposed to ship without the onboard memory controller. Meanwhile, the brit sites have noted a trend on low-end SMP boards:

http://theinquirer.net/?article=11845

...manufacturers are opting to use only one chip's controller. This clouds the issue a good bit, but maybe the impact can be derived somehow. (Gah, can't just take the hit versus 'regular' bank-per-CPU designs, because the data in the banks aren't coherent... Maybe you could try to derive the impact on regular bank-per-CPU multiprocessor boards versus uniprocessors, but that seems pretty specious, sounds like a job for synthetic benchmarking.)

http://tinyurl.com/oa40 pretty much stops at the edge of the L2, annoyingly. At least for a dork like me.


@Andreas:

Wow. I was looking for those numbers, and I think it's fair to say "Holy ****, how'd they **** that up?"

Barefeats suspects the cache is the story. I dunno.
http://tinyurl.com/pen1 links to Apple's sparse documentation of their "U2" chipset. The concept of "prefetch" comes to mind, but I'm not skilled enough to apply it, and my halfa__ed research suggests that'd be the wrong tree... Alternatively, it smells like buffering/"converting" from DDR down to MPX (this is MPX, right?) adds a latency they couldn't counter. (Hm, but putting more?/faster? SRAM in the controller, and then cranking up the memory even faster -- DDR400 vs. 333 -- and presumably the internal chipset operations along with it -- less wait to 'undouble' each bit, pack it into that SRAM, and get it aimed at the CPU -- should compensate? Maybe U2 isn't asynchronous *enough* internally, since Apple only did it for "marchitectural" reasons, and fear of the price spike for plain SDR?)

Since the few Articia benchmarks we have have been pretty close on both platforms, we have a good baseline for the existing hardware... Guess we'll have to wait and see if the Discovery III beats it. (OTOH, the numbers I'm thinking of -- http://personal.inet.fi/cool/pekosbil/a1benchmarks.htm -- were mostly tasks that could run out of cache? Anyone want to post some LAME numbers or something?)
Marvell announces Discovery III Northbridge : Comment 86 of 94ANN.lu
Posted by tarbos on 02-Oct-2003 04:34 GMT
In reply to Comment 85 (Joe "Floid" Kanowitz):
|Northbridge @2.2 Ghz FSB. >I know (or hope) you're just playing, but he meant the on-die memory controller. I see. It's just confusing since there is no "northbridge" per se on AMD64.The RAM controller at first was only good for DDR333 with Opteron andHypertransport also doesn't run at 2,2GHz - so why draw a (non)comparison hereto other systems' real northbridges/systemcontrollers?An onchip memory controller is nothing new, certain IBM and Motorola SoC PPC haveit now (also with DDR Ram support) - Transmeta even adds AGP.
Marvell announces Discovery III Northbridge : Comment 87 of 94ANN.lu
Posted by hammer on 02-Oct-2003 04:40 GMT
In reply to Comment 86 (tarbos):
>I see. It's just confusing since there is no "northbridge" per se on AMD64.
AMD has integrated most of the Northbridge functions on the CPU die.
Marvell announces Discovery III Northbridge : Comment 88 of 94ANN.lu
Posted by hammer on 02-Oct-2003 04:58 GMT
In reply to Comment 85 (Joe "Floid" Kanowitz):
>I'm really lost on roadmaps right now, but whatever ClawHammer was/is, it was
>once supposed to ship without the onboard memory controller.
The value end Athlon 64(a.k.a. Clawhammer)(Socket 754) has in-built memory controller (single channel). Unlike the Pentium 4, it’s a full duplex bus (i.e. 3.2 GB/s outflow and 3.2 GB/s inflow). AGP tunnelling has its own separate bus/bandwidth allocations (for AGP8X it's about ~2GB/s). Both the Athlon FX51(Socket 940*, Socket 939**)and Opteron(Socket 940*) has the onboard dual channel memory controllers.

*The same motherboard platform.
**For the cheaper 4 layer motherboard class and normal un-buffered DDR SDRAMs.
***Onboard memory controllers can be turned off, IF one desires to build the classic Athlon K7 way.
Marvell announces Discovery III Northbridge : Comment 89 of 94ANN.lu
Posted by hammer on 02-Oct-2003 05:25 GMT
In reply to Comment 87 (hammer):
Addendum...
Refer to this website for reference (in relation to K8 vs K7 vs PIV)
http://www.a1-electronics.net/AMD_Section/CPUs/Athlon_64+FX_Sept03.shtml
Marvell announces Discovery III Northbridge : Comment 90 of 94ANN.lu
Posted by tarbos on 02-Oct-2003 07:14 GMT
In reply to Comment 89 (hammer):
"AMD Athlon 64 processors use AMD's HyperTransport to interface with the North& South bridges instead of through the old FSB" So if this is to be the new "FSB"... >AMD64’s northbridge and CPU FSB runs at full speed with the CPU core ...there is no such a thing like "AMD64's northbridge" or CPU FSB running atCPU core speed since Hypertransport clocks at "800MHz DDR (or effective 1,600MHz)".Common PC northbridges implement AGP or Gigabit Ethernet in addition to the memorycontroller and an interface to the southbridge. Others have multiple PCI(-X) busesand other goodies that AMD64 CPU itself lacks. It just offers a fast (Hypertranport)interface and that's about it. You still need several bridge chips to make availablethe usual PC mainboard features so you cannot speak about the functionality thatmoved into the CPU itself as a full fledged northbridge seperated from the coreby a 2.2GHz FSB and therefore being inherently faster than everything else.But in the end everything may be a matter of definition, of course.
Marvell announces Discovery III Northbridge : Comment 91 of 94ANN.lu
Posted by Andreas Wolf on 02-Oct-2003 09:07 GMT
In reply to Comment 85 (Joe "Floid" Kanowitz):
> Since the few Articia benchmarks we have have been pretty close on both platforms,
> we have a good baseline for the existing hardware... Guess we'll have to wait and
> see if the Discovery III beats it. (OTOH, the numbers I'm thinking of --
> http://personal.inet.fi/cool/pekosbil/a1benchmarks.htm -- were mostly tasks that
> could run out of cache? Anyone want to post some LAME numbers or something?)

Those CPU benchmarks completely run within the cache so another northbridge wouldn't change a bit.
Marvell announces Discovery III Northbridge : Comment 92 of 94ANN.lu
Posted by hammer on 02-Oct-2003 20:17 GMT
In reply to Comment 90 (tarbos):
>Others have multiple PCI(-X) buses and other goodies that AMD64 CPU itself
>lacks. It just offers a fast (Hypertranport)interface and that's about it.
>You still need several bridge chips to make available the usual PC mainboard
>features so you cannot speak about the functionality that moved into the CPU
>itself as a full fledged northbridge seperated from the core.

The features mentioned can change from time to time BUT the critical elements
that makes up the classic Northbridge does not change. The other goodies
doesn't change the fundamental operation of the PC i.e. Without a memory
controller the desktop PC wouldn’t function. The Desktop PC can function without a AGP or PCI-Express GFX cards, plain PCI V2.2 would be sufficient.

>An onchip memory controller is nothing new, certain IBM and Motorola SoC PPC
>have it now (also with DDR Ram support) - Transmeta even adds AGP.

I didn’t say it was special i.e. where did you get this idea?
Anyway, none of the CPUs you mention runs at 2.2Ghz for critical Northbridge functions.

>The RAM controller at first was only good for DDR333 with Opteron and

Limited by the certified registered PC2700 DDR SDRAM standards,
one can overclock the memory controller to DDR400. Such as setup is
similar to the Athlon FX51. Stepping C0 shouldn’t oppose a problem for
both of the CPUs.

>I see. It's just confusing since there is no "Northbridge" per se on AMD64.

There are 3 critical elements that makes up the classic K7 Northbridge.
(e.g. VIA KT400/600 series, Nvidia NForce I/II series).
1. Memory controller. Interfaces with RAM modules**.
2. AGP bus**.
3. Northbridge<>Southbridge I/O**.

With AMD K8 processor.
1. Memory controller is integrated**.
2. AGP bus via integrated AGP tunnelling**.
3. Southbridge I/O Link via hyper-transport link.

This hasn’t change since the invention of Northbridge and Southbridge
architecture.

Anything else is just optional features i.e. they can go to Southbridge
(e.g. MCP-T*, MCP-S1000 Southbridge). *Gigabyte's NForceII 400 Ultra solution
(includes Gigabit Ethernet). Features can change from time to time
e.g. One doesn’t need AGP GFX card to get a display.

In the old K7 nForce II 400 Ultra...
(CPU e.g. 2.2Ghz) <CPU's FSB e.g. DDR400> (Northbridge Functions e.g. SPP)
<HT interconnect e.g. @800MB/s> (Southbridge e.g. MCP-T).

In the K8 platform...
(CPU core e.g. 2.2Ghz) <at CPU's speed e.g. 2.2Ghz> (Critical Northbridge Functions**)
<interconnect i.e. 800MHz DDR> (Residual/optional**** Northbridge functions/Southbridge).

****Features such as Gigabit Ethernet wouldn't be considered as a critical
function to the operation of the PC.
Marvell announces Discovery III Northbridge : Comment 93 of 94ANN.lu
Posted by hammer on 02-Oct-2003 20:50 GMT
In reply to Comment 92 (hammer):
Addendum...
> Anything else is just optional features i.e. they can go to Southbridge.

VIA VT8237 has its Gigabit Ethernet connected to the Southbridge (use in VIA's KT600 Apollo, K8T800 and PT800 product lines).
Marvell announces Discovery III Northbridge : Comment 94 of 94ANN.lu
Posted by hammer on 06-Oct-2003 21:20 GMT
In reply to Comment 92 (hammer):
Addendum

Reference from the official AMD Inc.

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_9485_9488^9494,00.html

“Integrated Northbridge | Yes, 128-bit data path @ CPU core frequency” - AMD
Anonymous, there are 94 items in your selection (but only 44 shown due to limitation) [1 - 50] [51 - 94]
Back to Top