what's that one quote from Cordelia from Buffy the Vampire Slayer?

"I don't buy stuff cuz its more expensive, but because it costs more."

In our next Tech Bits, Tim will talk about how he spent $1000 on a P4 Extreme Edition proc. Because if it cost more, IT MUST BE BETTER.





May 3rd, 2005 - "We are preparing to destroy you." http://www.nin.com

You pricks think that you're so well informed. PC-1066 is 3 years old. There has been 1200MHz, 1333MHz and 1600MHz memory chips since then. All with stellar performance.
http://www.anandtech.com/cpuchipsets/showdoc.html?i=1960

...err yeah! From that page, it sure sounds like Anandtech is beside themselves with glee that a bunch of trend-setters like SiS are still on the RAMBUS crackpipe:

For whatever reason, SiS will continue their RDRAM support throughout 2004 and into 2005 with these two chipsets. Other than the memory interface, everything else remains the same as the DDR based chipsets.

-----
MeatFarts

You pricks think that you're so well informed. PC-1066 is 3 years old. There has been 1200MHz, 1333MHz and 1600MHz memory chips since then. All with stellar performance.

http://www.anandtech.com/cpuchipsets/showdoc.html?i=1960


I can't wait for a real eugenics movement, i'm bored by short people.

You sir, are an asshat.

Honestly, you spend all this time trolling for Rambus "to increase [your] total number of posts." Just to wrap this up, here are some real world (IE game tests) benchmarks of rambus vs. DDR. This was the most recent iteration of the technology. They prove (in my mind, at least) that although Rambus had significant throughput advantages their higher latency created a bottleneck, that ultimately lead to less than stellar performance.

http://www6.tomshardware.com/motherboard/20030331/sis658-09.html

Regardless of performance, Rambus's business tactics (a closed standard, sueing DDR (an open standard), over charging for their product and specifically the aforementioned "2 stick" requirement) place them square in the tech gutter. Rambus deserved what it got, hopefully these suits will be dismissed and we'll finally see the last of them..

ok would you say Processor system buses of modern Intel Pentium 4 CPUs are capable of pumping up to about 8.40GB data per second?. Quad-channel RDRAM memory can theoretically provide us 9.60GB/s, 10.66GB/s or 12.80GB/s throughput when using 1200MHz, 1333MHz or 1600MHz memory chips.

Obviously, such memory bandwidth is not something desktop Pentium 4 systems really need.

So what's the point?

after the 4th post I just wanted to increase my total number of posts.


This comment was edited on Jan 26, 18:58.

My statement was accurate, arguing semantics now are we?

No. Your statment was false.

From your first post:

RDRAM was the superior memory technology

It wasn't and still isn't.

Also:

If the market had embraced RDRAM without a doubt computer performance would be greater today.

That's unsubstantiated, based purely on the minor nit that RAMBUS hasn't significantly better when it was available. And by significant I mean exactly that -- statistically significant. The range that it outperformed was never outside the "white noise" category.

Why are you so threatened of the idea of them coexisting?

I'm not. RAMBUS is. Which is why they angled for an exclusive deal with Intel and why they're suing everyone under the sun, regardless of merit.

Are you denying that if RDRAM technology had been pursued that it would be worse off?

That sentence doesn't parse. Even in context. I can guess what you meant, however, and I question that it would be better off. You seem to like talking about competition, except that you neglect to mention that RDRAM is the antithesis of that. It's controlled entirely by a single company who could (and did) choose who to license it to and who not to. Monopoly doesn't exactly beget innovation.

Ok, I'm done. You've continued to ignore the facts. You don't appear to know jack shit about computer design. You certainly don't know anything about memory latency vs bandwidth and the relative impacts of each on a modern general purpose computer. It's clear that you aren't going to be swayed by logic, facts, or reality. So what's the point?

So, Tim... has the fact that all of your neighbours likely run filthy filthy DDR (or even *gasp* SDR) RAM lowered real-estate values in your area, or do you live in one of those gated RDRAM communities?

You do gotta be careful, or you might get STDram..

_____
Enahs
That’s a deep kiss too, like the Europeans. You know the French they have to unhinge their jaw to show love.

Fuck the average consumer.

I can't figure out for the life of me why all of these people are so dead set against Rambus. I guess it's like....the majority of the unlearned masses are lemmings who followed the sometimes twisted reviews, and joined in the chorus against Rambus. More expensive...HA! Compare prices on your Corsair stick.
My guess is that all of these little bugs who started this anti-Rambus campaign were working at McDonalds when Rambus first appeared, and the hope of buying such an advanced piece of equipment was beyond their McBudget, and that hate is still alive.
DDR may have won simply from a corporate standpoint and Intel forced to abandon RDRAM, due to the media's manipulation of the unenlightened masses.

The throngs wanted an inferior memory solution because they are worried about dollar differences and they didn't want the inventor of the better solution to make his 10 dollars.

I think that says it.

Hey WOW! Real-life RAM snobbery!

So, Tim... has the fact that all of your neighbours likely run filthy filthy DDR (or even *gasp* SDR) RAM lowered real-estate values in your area, or do you live in one of those gated RDRAM communities?

-----
MeatFarts

My statement was accurate, arguing semantics now are we?

I’m talking performance, not value.

Who is losing? I am, unlike you I only can speak for myself. I’m losing because RDRAM promised a greater progressive future of performance. But more importantly it gave me a choice and an option. Mcbudget DDR owners could not have a higher performing memory looming out of there petty price range...so it had to go. Why are you so threatened of the idea of them coexisting? Are you denying that if RDRAM technology had been pursued that it would be worse off?


I can't wait for a real eugenics movement, i'm bored by short people.

I can't figure out for the life of me why all of these people are so dead set against Rambus

Because we can do math and have actually taken CS, EE, and Computer Engineering courses? Not to mention actually using that knowledge.

Naaah.

The throngs wanted an inferior memory solution because they are worried about dollar differences and they didn't want the inventor of the better solution to make his 10 dollars.

Even presuming your statement was accurate, which it just simply isn't, if you increase overall system cost by 10% while only increasing system speed by 5% then you don't have the better solution (presuming there is an alternative which can increase system speed by 5% at a lower cost... which there was).

RIMM 4200 is 32-bit RDRAM module

Ah, I sit corrected. Bloody stupid naming, not that that's new. Basically RIMM4200 is the exact same thing as RDRAM 1066, but it's 32 bits wide instead of 16. Essentially dual channel memory on a single stick of RAM.

Of course, we've already gone over the pointlessness of dual-channel setups.

What? You doubt? Hell. We'll even use the page you gave to show the "superiority" of RIMM4200.

Take a look at the benchmarks. Note the speed differences in every single benchmark between the RIMM4200 system and the PC-1066 RDRAM system.

Memory bandwidth: 1.5%
PCMark 2002: 7.6%
3DMark 2001: 0.98%
Comanche 4: 1.9%
Quake3 demo 4: 1.3%
JK2: 1.8%
Photoshop 7: 0.46%
Media Encode: 0.58%

Yeah, those are some whomping huge differences between the PC-1066 RDRAM system and the RIMM4200 RDRAM system. In fact, virtually all of them are so small as to be well within the margin of error. So your claim that the comparison against a 16-bit RDRAM module not being equal is, well, crap.

Keep deluding yourself into thinking that RDRAM is in some fashion, superior. It isn't. Keep thinking that the marketplace made a bad decision. It didn't. The increase in system cost for RDRAM system vastly exceeded the increase in speed. Particularly when the speed improvement was negative.

Fuck the average consumer.

I can't figure out for the life of me why all of these people are so dead set against Rambus. I guess it's like....the majority of the unlearned masses are lemmings who followed the sometimes twisted reviews, and joined in the chorus against Rambus. More expensive...HA! Compare prices on your Corsair stick.
My guess is that all of these little bugs who started this anti-Rambus campaign were working at McDonalds when Rambus first appeared, and the hope of buying such an advanced piece of equipment was beyond their McBudget, and that hate is still alive.
DDR may have won simply from a corporate standpoint and Intel forced to abandon RDRAM, due to the media's manipulation of the unenlightened masses.

The throngs wanted an inferior memory solution because they are worried about dollar differences and they didn't want the inventor of the better solution to make his 10 dollars.

I think that says it.


I can't wait for a real eugenics movement, i'm bored by short people.

"If one could afford the best why wouldnt, they buy it?"

what if they don't need the best? You're making two assumptions, that the average consumer can afford it, and that RAMBUS is the best, but I'll leave that to others to disprove.

Its the business's goal to make a product we WANT to buy, if they can't do that, then they failed, *NOT* the consumer, pure and simple, regardless of how you try to spin it.

Sorry, RIMM 4200 == RDRAM 1066.

RIMM 4200 is 32-bit RDRAM module
The article you linked showed 16-bit RDRAM module hardly 'equal'.

http://www.sis.com/products/chipsets/oa/pentium4/r659.htm
A chipset form early 2003 with RDRAM, how does that compare to DDR chipset of today?

http://www.ocztechnology.com/aboutocz/press/2003/39

If RAMBUS doesn't make a product that we want to buy, how is that our fault?

If one could afford the best why wouldnt, they buy it?

This comment was edited on Jan 26, 15:50.

"nearsighted consumers"

anyone else think that Tim is shooting himself in the foot with that arguement?

If RAMBUS doesn't make a product that we want to buy, how is that our fault?

*ANY* business that blames the consumer obviously doesn't deserve to be in business

That’s a dandy article but this article doesn’t have anything to do with RIMM 4200.

Sorry, RIMM 4200 == RDRAM 1066. It's not my fault that they couldn't decide on a single moniker for the exact same damn thing.

If you'd like to go over the fine details of latency vs bandwidth, go read the articles about it on Ars Technica. They did an excellent job. Then maybe you'll understand why there's virtually no difference in overall system speed between single and dual channel architectures, why increasing front side bus speeds gives a minimal system speed improvement, and why RDRAM didn't perform as advertised.

say "NO!" to software patents!

http://lpf.ai.mit.edu/Patents/patents.html

http://www.newscientist.com/article.ns?id=dn2178

http://www.nosoftwarepatents.com/

Now, that said, here's a site that's actually well known for benchmarking, and generally well respected:
http://www.extremetech.com/article2/0,1558,590673,00.asp

That’s a dandy article but this article doesn’t have anything to do with RIMM 4200.
Find one.


RDRAM was and is more suitable for the future from a technological stand point. When RDRAM launched it was faster then existing memory technology and it was in its infancy. The competing RAM was near the end of its cycle with no where to go technologically speaking. Now its over and where do they go now?


I can't wait for a real eugenics movement, i'm bored by short people.

RIMM 4200 the last RDRAM to hit the market vs. what was competing with it at the time.
http://www.gamepc.com/labs/view_content.asp?id=ngrdram&page=6

Wow! GamePC -- there's a name I go to for benchmarking. Woohoo! And on top of that, they benchmarked the top of the line Intel boards with top of the line RDRAM against... yeah... SiS! Always known as a top notch performer! Oh, and look! No real world benchmarks! Just synthetics. My my my.

Memory benchmarks, if you haven't figured this out yet, are pretty useless. A dual channel system will show memory throughput benches of nearly twice a single channel system -- as it should. Start running actual programs on both systems, however, and you'll see <4% difference between the two because bandwidth just doesn't matter.

Now, that said, here's a site that's actually well known for benchmarking, and generally well respected:
http://www.extremetech.com/article2/0,1558,590673,00.asp

I believe Tech Report and Ars Technica have similar articles, but Tech Report doesn't play nice with Flashblock (known bug in Firefox), so I'm not going to try and find the appropriate articles.