Intel X48 Motherboards to Show Up This Week

Kougar

Techgage Staff
Staff member
It's kinda ironic... memory prices for a 2GB DDR2-800 kit at the start of 2007 are the same as some good 2GB DDR3-1600 kits right now.

The irony is that at the time $200 was considered a good price for 2GB DDR2 last year. ;)
 

sbrehm72255

Tech Monkey
Yup, times are a changing, I was one of the big holdouts in switching over to DDR2 from the old DDR (mainly due to running AMD A64 X2's).

I only switched over because of the Intel C2D's and now I'm up running DDR3 (low end stuff) early.
 

Rob Williams

Editor-in-Chief
Staff member
Moderator
I was pondering about the the other night... I was thinking back to just two years ago when I was having a blast overclocking the Corsair 2GB 3500LL-PRO kit... amazed to high-hell that I could reach a stable OC of DDR-540 with 3-3-2-0 timings...

It felt like less than a month later, DDR2 was scattered all about here. The transition was SO fast, it was incredible. Though even at that time, DDR2 was around, but everyone had AMD... I jumped on the DDR2 bandwagon a little earlier than most only because I knew there would be demand, and demand, there certainly was.

Now we are dealing with DDR3, and I have to say, it bores the hell out of me. I just don't find overclocking DDR3 to be as fun or interesting as I did both DDR/DDR2... I am not sure why. Perhaps a part of it is that we saw DDR3-2000 overclocks AT LAUNCH, or it could be that technology is just moving so fast, it's hard to enjoy focusing on one architecture for too long.

Before we know it, DDR4 will be here.
 

Kougar

Techgage Staff
Staff member
Well, it would help if DDR3 showed some appreciabe gains in everyday applications, instead of narrow situations or specfic benchmarks. DDR2 had an easier time of doing that once it began reaching 800Mhz with around CAS 4 latencies... DDR3 even at 2GHz just doesn't do much.

I am rather curious how Nehalem will change this though, removing the FSB latency, integrating the memory controller directly... I am betting it will to some degree. Just how much I wish I knew! :D

Speaking of DDR4, you did hear that ATI was playing around with Samsung GDDR5 right? Plain nuts... :)
 
Last edited:

sbrehm72255

Tech Monkey
GDR5 is on its way (maybe, depending on price I'd bet)

I'm willing to bet that the onboard controller is going to make a big differance, sort of what was/is going on between AMD and Intel systems running DDR, AMD always won the bandwidth race with their onboard controllers.
 

Rob Williams

Editor-in-Chief
Staff member
Moderator
Both GDDR5 and DDR4 are en route, but not necessarily right away. I know that JEDEC was in close contact with memory manufacturers regarding DDR4 last spring... before DDR3's launch. Talks and development will go on for quite a while before something gets pushed out.

New memory types are always expensive as hell (obviously), so I don't think we'll see GDDR5 that soon.

AMD always won the bandwidth race with their onboard controllers.

Yup, and with Intel's QuickPath Interconnect and the three-channel memory configurations, bandwidth and latencies should be the best we've ever seen.
 

sbrehm72255

Tech Monkey
Yup, and with Intel's QuickPath Interconnect and the three-channel memory configurations, bandwidth and latencies should be the best we've ever seen.

That's exactly what I'm thinking and hoping for anyways,,,,,,,,,,,,:D

It's going to put all that more pressure on AMD to pull their heads out and do something fantastic.

I know that this info is from the Inquire, but it's still good info none the less..................;)

Analysis Nehalem's happy memories: record DRAM bandwidth

Supported speeds
Officially, at this moment, Bloomfield and Gainestown will support up to DDR3-1333 memory speed in both standard and registered varieties (servers only for the latter), both ECC and non-ECC. Does this sound sluggish, knowing that this CeBIT saw OCZ, Corsair, Kingston and others show DDR3-2133 memories, running at CL9 no less, on Nforce 790i?

Not really - the first DDR3-1333 CL5 ultra low-latency parts are out, some of them in 2 GB density per DIMM module. On the desktop Bloomfield, with three channels and two DIMMs per channel, this gives you a comfortable - even for Vista bloatware - 12 GB RAM. The bandwidth? A whopping 32 GB/s total peak memory bandwidth across all channels, same as dual-channel DDR3-2000.

Furthermore, there is no FSB bottleneck that lets current Penryns access only half of that huge dual-channel DDR3 - the three channels in Nehalem are, through the on-board memory controller, accessed directly by the CPU's L3 cache. Even if Intel just bolted on a 3-channel version of X48 chipset's DDR3 controller, and removed the FSB blockage, you'd still be able to get at least 25 GB/s Sandra memory bandwidth scored with such DDR3-1333 memory configuration.

What about the dual socket part, the one to power the Skulltrail successor? Well, Intel's slides show it having slightly over four times the bandwidth of the best current dual-socket Harpertown with four FBD-800 channels. Knowing we easily get 9 GB/s there, we're talking about a 36 GB/s dual-socket workstation Sandra memory bandwidth score - a massive leap of nearly three times over the best AMD systems too.

http://www.theinquirer.net/gb/inquirer/news/2008/03/20/analysis-nehalems-happy
 
Last edited:

Kougar

Techgage Staff
Staff member
Nice link Bio, I had missed that article!

With that much bandwidth, my theory about Nehalem being more sensitive than Penryn to memory speeds might be wrong. It sounds almost like Nehalem will have so much bandwidth with its direct low latency triple pipe to the memory that faster memory would be superfluous. I still bet that low latency RAM will show some nice improvements though!

Does anyone remember what ultra low CAS latency it was that AMD's chips couldn't make use of? Do they even still have any ultra low CAS limitations? Have to briefly wonder if Nehalem might share a similar issue.
 
Last edited:

Rob Williams

Editor-in-Chief
Staff member
Moderator
It was AM2 that had the limitation... none of the chipsets would allow a CAS latency of 2. However, many people were overclocking their modules so high that CAS2 wouldn't have been possible anyway, so I didn't see a huge issue there.

As for that Inquirer article... I had no idea DDR3-1333 C5 modules were available... that is damn sweet. I need to get back in the loop.
 

Rob Williams

Editor-in-Chief
Staff member
Moderator
I find it interesting that the kit there runs at DDR3-1375 with CL5... but requires tRC of 7. With DDR2... I found the CAS latency to be far more finicky than the tRC.

I am impressed that CL5 kits are available regardless though... I am really out of the loop. I should get a bunch of these kits in and see how they perform to the first-gen.
 
Top