Slashdot Mirror
Is Rambus Destined to Return?
An anonymous reader pointed us to an article running over at Tom's that talks
about the world
of ram and criticizes the performance of DDR. The article goes
into DDR333, DDR400, and Rambus, and explains the issues at higher
clockspeeds.
It seems to me that Rambus has offended so much of the industry that it even intel's continued (though lately lessening) support, or perhaps especially with intel's support it will fail to be implemented by the majority of m/b manufacturers.
Other avenues for gaining speed exist- like Nvidia's extra memory controller for the gpu in the xbox and higher end nForce chipset.
The fact is that most IC testers today support the lower-speed parallel connections. High-speed serial connections like Rambus and SERDES require very expensive mixed-signal testers with expensive and complicated load boards (the PCB between the tester itself and the chip). These high-speed serial I/Os on the memory ICs themselves are also generally much larger than on a DRAM, probably by a factor of 5. So, you don't get die savings, you don't get lower test costs, and most of all you don't have any processors whose front-side buses exploit this. Plus, you have very expensive target products in terms of motherboards to support the Rambus ram requiring tight trace routing and signal isolation, and their very limiting 28ohm max impedance (at least with the PC800 RDRAM), almost completely opposite in difficulty to DDR. So where's the advantage?
If you also figure that the memory controllers for Rambus are configured for dual-channel operation, it becomes much clearer that the advantage is not in the memory architecture itself but in the controllers. Suppose a server board manufacturer decides to support quad-channel PC2700 1GBx4. That's 10.8 GB/s of potential memory bandwith on sequential accesses! There's hope with chipsets like the Nvidia nForce420 dual-channel DDR, but the Athlon FSB is the limiting factor there. And let's not get into the infamous first-access latency issues which I hope they're finally addressing.
Rambus is also notorious for poor tech support. I worked for a major silicon vendor using their core, and they never responded to our requests for minimum PLL-to-Rambus core distances. It was abjectly ridiculous, but not surprising considering that regular SDR/DDR memory interfaces outnumbered Rambus designs 100:1. Have things changed? Considering what their legal bills have been lately and an erosion of their tech support, I doubt they can afford to improve it much.
IS there anything else in memory arch besides rambus and DDR on the horizon?
An Education is the Font of All Liberty
I think the point of that article is not that the Athlon is going to burn up.
There was a rumor being spread by some kids with AMD processors that the Pentium 4 runs at half speed whenever you do more than just checking email with your computer. They had taken a new feature in the Intel processor and manipulated it into a fault. Tom's article was only trying to explain what the feature is and why it's good, since many people did not understand what it was about. He was not in any way saying that your heatsinks are going to fall off and your processor will burn up, if you use an Athlon processor. It was not an attack on AMD, it was just an explaination of a new feature found in certain other processors.
About a year or so ago, intel was doing demo's of Q3A at fry's to show off the new P4. Economy was good, I had like 4k from PTO that I aquired between jobs so I said what the hell and dropped $1500 on a 850dgb board, processor, case and 128 of ram.
I gotta say, this stuff is hot, my friends have all gone off and bought gforce3's, amd's with DDR. I thought these new cards/systems would have score winmarks well above my own (around 3800 with a gforce2gts) but I was surprised to see they only score 1000 or so more than me.
Out of curiosity, we put one of those GF3's in my system. Without fail I would score about 400 to come in around 6300 3dmarks above my buddies amd1.6. My P4 is just 1.4. Yet even with a lower clockrate the memory bandwidth made a huge difference.
I'm not trying to cause a ruckus here, anyone with deep enough pockets (or access to enough systems) can just as easily do the same testing I did. Bottom line whether or not the moderators like it is rambus systems do provide the absolute best possible performance in 3D gaming. It certainly was expensive when it came out but now with the falling prices of all ram, it's within reach of anyone that want's that extra "oomph" in thier system.
Does anyone know of any AMD boards that use rambus? I'm sorta curious what kind of scores those get in comparison to the intel one's. Anyways thats my comment.
The days of cheap memory are over.
They say this because of the huge expense needed to provide 512MB or more of ultra fast memory. But what if they added yet another level of "cache"?
Put in 128MB or more of super-fast RAM (faster than today's RDRAM or DDRAM, maybe using an exotic bus) backed by gigs of cheap, easy-to-make memory (PC266 DDRAM or slower). The cheap ram is still orders of magnitude faster than a disk drive. Manage them with hardware that does page swapping similar to virtual memory.
You could get good system performance and lower overall cost.
After Tom's Hardware printed that article where all they glossed over the PS2 and Gamecube as if they were useless (and all the information about those 2 consoles seems to have come straight from Microsoft) and then devoted 12 pages to details about the Xbox, I'll never really believe what they write about products again. I'm not even really sure if I believe all their AMD benchmarks, anymore. They were clearly bought by MS for that, so they probably were bought by AMD for those other articles.