posted
I guess, it's coz the official JEDEC standard is DDR400... so anything above that speed is just not approved. And they want to sell DDR2, so nobody would waste time for a fresh new JEDEC standard for simple DDR
-------------------- cheers! Ivo Core Duo E6300 @ 3,2GHz|Gigabyte GA-965P-DS4 rev.3.3 |4x1024 MB PC6400 OCI @ 914mhz|Leadtek WinFast PX8800 GTS 640MB|Raid0=3x160GB Samsung SP1614C|Zern PQ plus Waterblock Posts: 1470 | From: Germany/Bulgaria | Registered: Feb 2003
posted
Taking a quick look at Crucial's website - their stock ddr2-5300 (667) ram has cl=5 at 1.8v; the high performance ballistix ddr2-5300 (667) ram has 4-4-4-10 timings at 1.9v.
Not exactly latencies to write home to mom and brag about..
-------------------- RHCF Rosetta challenge: Crunch faster than my single default speed 24/7 machine and put me in last place for the team. Help RHCF, prove your O/C abilities, run a Rosetta client or more. I dare you.. Posts: 1179 | From: Kodiak, Alaska | Registered: Jul 2003
posted
Funny! I just stumbled across a DDR2 thread at another forum:
quote: OK,I've up the Memory voltage from what I presume the default of 1.8v to 1.9v and according to cpu-z v1.28.4 - its 4-4-4-12. Don't laugh ! at least its improved from the older 4-4-4-14 ! just ran stability tests with 3DMark03,05 and PCMark04 and hey, I got a little improvement with PCMark04.
Imagine? 4-4-4-12....
-------------------- Too Many Computers,... Too Little Time ..... Damn Phenoms! Posts: 25191 | From: Fire Island, NY | Registered: Feb 2003
Each of the above seems to have slightly different tidbits about DDR2 - while the next 3 are mostly just reviews/tests on single low latency ddr2 modules:
So it looks like the ddr2 latency penalty is quickly disappearing..
-------------------- RHCF Rosetta challenge: Crunch faster than my single default speed 24/7 machine and put me in last place for the team. Help RHCF, prove your O/C abilities, run a Rosetta client or more. I dare you.. Posts: 1179 | From: Kodiak, Alaska | Registered: Jul 2003
XDR as lined out for the Cell Processor (anandtech):
quote: dual channel XDR memory controller, each channel being 36-bits wide (32-bits with ECC). Cell’s XDR memory bus runs at 400MHz, but XDR memory transfers data at 8 times the memory bus clock - meaning that you get 3.2GHz data signaling rates. The end result is GPU-like memory bandwidth of 25.6GB/s.
Maybe AMD will one-up intel and go right for something crazy like XDR. Due to AMD's smaller market share, pricing would probably be an issue for dimms. BUT, the chips themselfs should be cheap. If these chips are to be used in a mass-produced game console, it should be even better if a PC market developed.
Intel must still have relations with Rambus. They messed up with them once, how ironic would it be for Intels little competetor to cozy up next to Rambus, and exceed where they failed?
Neither DDR3 or XDR have been released yet, so its up in the air as far as that. Using XDR would require licencing, and that in itself might keep AMD away from XDR.
I put my chips on XDR for which platform i would like to see. Both DDR3 and XDR would require new memory controllers, new sockets, and new chips. Who's to say what they will use.
-Plack
-------------------- Lap: Dell Vostro 1500/1.6GHz C2D/4GB RAM/Nvidia 8400GS 128MB/15.4" Desk: E8400 9x433(3.9GHz)@1.35v/Gigabyte G33M-DSR2/4GB RAM/ATi 3850 256MB/37" Westinghouse 1920x1080 My Photo works on Deviant Art Posts: 1313 | From: Hephzibah, GA | Registered: Jan 2004
Somewhere along the line of this overload of urls today, I've run across mention that DDR2 is double the data rate of DDR - and that 4-4-4-8 DDR2 800 would have the same basic latency as 2-2-2-5 DDR 400 (pc3200); only the DDR2 part would have twice the bandwidth. DDR3 would have twice the bandwidth and clock of the DDR2 part (4 times DDR) so a 8-8-8-20 DDR3 1600 part would have the same general latency as a 2-2-2-5 DDR 400 (pc3200) part; having 4 times the bandwidth of the DDR part.
Being as how DDR has a 10-12.5ns latency - and the claims that XDR has a 1.x to 3.x ns latency - I'll be opting for it as well; especially if Rambus has found a way around the major problems that destroyed the chances for Rambus the first time around.. 1. latency much higher than DDR - hmm.. that's been taken care of. 2. Outrageous pricing to customers since Rambus was too greedy - and outrageous pricing to customers because few were willing to pay for the license since.. Rambus was too greedy. 3. Engineering problems causing delays that caused the parts to arrive with no advantage over the then current DDR parts. (If Toshiba is prototyping 'em.. then this probably isn't going to be a problem this time around..)
But back to Polygon's question - with DDR's pc3200 ram being considered budget now - it's strange we haven't seen faster speed grades designated as the "default" ram speed for the newer systems.
-------------------- RHCF Rosetta challenge: Crunch faster than my single default speed 24/7 machine and put me in last place for the team. Help RHCF, prove your O/C abilities, run a Rosetta client or more. I dare you.. Posts: 1179 | From: Kodiak, Alaska | Registered: Jul 2003
quote: FlexPhase will allow precise on-chip alignment of data with clock. This doesn’t mean as much to consumers as it does motherboard manufactures. The manufactures don’t need to be worried about PCB trace lengths matching and PCB timing constraints.
Easier motherboard design? Cheaper? Wow, this isnt the rambus we are used to! Easier RAM layout will remove the last weak link in motherboard design in an AMD64 system. AMD should know better than to build a CPU with a weak memory controller this time around. And if the memory traces are as easy as pie, there shoudl be some exellent memory supoort and overclocks right out of the box!
Man i hope AMD says something soon...
-Plack
-------------------- Lap: Dell Vostro 1500/1.6GHz C2D/4GB RAM/Nvidia 8400GS 128MB/15.4" Desk: E8400 9x433(3.9GHz)@1.35v/Gigabyte G33M-DSR2/4GB RAM/ATi 3850 256MB/37" Westinghouse 1920x1080 My Photo works on Deviant Art Posts: 1313 | From: Hephzibah, GA | Registered: Jan 2004
posted
The reason is quite simple on AMD systems. The Chipset has nothing to do with the memory anymore. The memory-standard is given by the used memory controller and here it´s inside the CPU.
They simply can´t announce the NF4 with something like "supports DDR 600" because this would not be true.
If AMD ever decided to support faster memory types like DDR 600 and relases CPU samples with this support, all the Mobos out there can be called as "ones that support DDR 600 in combination with such a CPU stepping", no matter they really can do it.
-------------------- remember... if the world didn't suck, we all would fall off Posts: 1328 | From: Germany | Registered: Feb 2003
Damn Phenoms!
![[Roll Eyes]](rolleyes.gif)
![[dono]](graemlins/fragend013.gif)
![[Banana]](graemlins/BANANA.GIF)
"
