Slashdot Mirror
Intel To Drop Rambus Exclusivity, Support SDRAM
Overload128 writes "Over on ZDNet, there's an article detailing Intel's plan to let Rambus memory stand on its own. It seems that it will stop bundling RDRAM with Pentium 4s and stop giving rebates to OEMs that use P4s and RDRAM. They are also releasing a new chipset soon, called 845 that will support SDRAM (SDR) with DDR support for P4s not far down the line."
Actually, I think AMD systems rock - I recommend them to my friends, and I have two of them at home.
The problem with AMD boxes in a corporate envirnment is consistency. When I buy a box for work, I'm not looking for blazing speed and lowest cost. I want a PC that'll be available in exactly the same config for at least 9-15 months from its date of introduction. There aren't any AMD-based systems being sold by the major manufacturers that meet that criteria - the typical "corporate" system (like the Dell Optiplex, Gateway E-series, etc.) will use a known stable chipset with a long planned lifecycle in the roadmap, Intel processors, and typically an ATI video chipset. When I buy them (in bulk), it's easy for me to use a standard system image to build the PC. Then all we have to do is generate a new NT SID, and set up the user profile. It only takes about a half hour for us to build a new PC as a result, including the time to image the PC.
As for the current P4 systems, the reason we're avoiding them is RDRAM. I have no interest in stocking multiple RAM types, and the performance isn't sufficiently optimized to make it worthwhile. For my purposes, there's nothing wrong with the P3/i815 platform (or BX, for that matter), but BX is gone and i815 will be rapidly phased out once i845/Brookdale is available around the beginning of October. Hotrods don't really benefit us at all (our current main platform is still 440BX/P3-650), but stability does. Brookdale is slated to be around for a couple of years, with the DDR version coming out early next year.
One other advantage of buying the "corporate" class systems is that all the manufacturers who want the business will cheerfully share their product roadmaps with you - including model names/numbers, availability dates, length of product cycles, and pricing. However, it's NDA'd, so I don't get to share, unfortunately. However, I should have i845 sample systems in here during August from a couple of vendors to test.
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
Rambus RAM is nowadays well under $1/MB (the Chip Merchant, for instance, sells 256MB of PC800 RDRAM for $170) - the only thing that makes it look expensive now is the utter collapse of the DRAM market. About six months ago, that would have been a reasonable price for PC133 SDRAM.
And Rambus has some technical strengths, too, when matched with a chip design like the P4 - the max bandwidth of RDRAM is higher than equivalent (PC133 DDR) SDRAM, though latency is higher. THe problem with Rambus was twofold:
First, the horrible business practices of Rambus, the company.
Secondly, when Rambus finally became available to the PC market (with the i820 chipset), the platform was so underwhelming that Rambus was effectively squashed. That was when the company turned their attention to litigating rather than working to improve the product - and we all have seen examples of companies that lead through legal agtion. They die.
And now that Intel is going to ship Brookdale, they might start selling some P4 chips at last. I know that, at least at my shop, we've held off entirely on PC purchases this year (except for a couple of servers and laptops), in order to wait for a viable P4 platform. I'm sure I'm not the only one - expect Intel's sales numbers to start rising again and some of the top-tier PC vendors to show signs of breaking out of their slumps. I, for one, have about 100 PCs to buy once there's a product worth buying.
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
The Rambus folks have single handedly run one of the most vile business models around. Regardless of the quality of their product (which thankfully is low) its important to make an active stand against this kind of behavior, because in the end it hurts us all. Imagine if a company like this had got some patents on underlying internet protocals. The internet would cost a ton and barely exist.
The opinion is widespread because people associate the MB chipset with the proc. And AMD 3rd party chipsets were (and still continue to be even today) notorious for shoddy hardware and drivers.
My brothers K5 was a flaky machine. My friend's K6-2 decided to just die. He always had sound card compatibility problems. Many IRC friends had sound issues with the early K7s. Video cards were always iffy ("Use the latest drivers! Upgrade every week!").
Notice that I left out any mention of the Motherboard chipset manufacturer. Most people, even techies, don't make the proper distinction.
A friend who works for AMD has a nice 1Ghz Athlon system. In Win2k it will occasionally hiccup for about a second or so. The problem? The VIA drivers suck. It's like building a great house but letting someone do a crappy job on your driveway and lawn.
Now, of course, some people haven't had any problems or researched the issues ultra-carefully before diving into the Kx fray but those people are few. I'm happy to say that a friend and I put together a nice Athlon rig for colocation and it's been doing just fine. However, this is an issue that has plagued AMD's public perception from the K5 days and that's why there is this reputation, deserved or not. I myself tread very carfeully when dealing with MB purchases.
Fsck cluebie moderators. I'll say what I want, offtopic or not. And fsck having to qualify every bloody statement just
The comment from the submitter makes it seem as if this is something new and important. This is actually really old news. Intel's 845 (Brookdale) chipset, which supports regular SDRAM and DDR has been in the works and well known for a while now and even benchmarked. Intel is probably initially only releasing it in the SDRAM flavor because of exclusive contracts with Rambus Inc. It is expected that in less than a year the DDR version will be out. Intel publically stated they are less than pleased with Rambus Inc. a long time ago.
As another poster mentioned the performance of the P4 with SDRAM is terrible. This is because the P4 was designed for memory with high bandwidth such as Rambus RAM and DDR as opposed to regular SD RAM. Tom's Hardware, perhaps the foremost Rambus hater, has an article on the 845 chipset and its very poor performance with SDRAM. Ace's Harware also has a summary.
All and all Intel's relationship with Rambus and use of Rambus RAM has been very foolish. The P-III was not designed to take advantage of the high bandwidth of Rambus so the improvements versus SDRAM were limited and the price of Rambus made VIA's competing SDRAM chipsets and AMD's solutions much more attractive. Now that Intel finally has a chip (P4) for which it makes sense to use Rambus RAM, Intel is slowly moving toward abandoning Rambus probably in favor of DDR. Although, given how hated Rambus is among RAM makers and the continued superior price-performance of DDR RAM, Intel's moving away from Rambus makes a lot of sense.
Am I the only one who still sees "DDR RAM" and thinks "dance dance revolution random access memory?"
so do you think my butterfly upswing another trickstyle will really push my computer THAT much faster?
no
At first this might seem like a good move, until you remember the only current selling point of the P4 is memory bandwidth (until SSE2 is mainstream). So effectively, the took the best feature and killed it to save on costs to the consumer. Too bad most consumers will see "1.8ghz now with 128 megs of SDRAM!" and buy it because they heard SDRAM is cheaper than RDRAM, yet the 1.8gig CPU is over $500.
--------
--------
It's OK to be social, just don't tell anyone about it.
Well, WinXP is just around the corner.
Not everyone deserves a 320i
I don't recall which Intel executive (one of the chief architects of the Pentium IV architecture) it was that lead to this common misconception, but he later refuted that this was what he meant.
This isn't a common misconception... it is, in essence, true.
The P4, as it stands today, has very small on-chip caches, certainly by comparison with, say, the Athlon, and the primary reason why that is is because of memory latency. Smaller caches have lower access latencies, mitigating the effects of RDRAM's high access latency. RDRAM also has very high bandwidth, significantly higher than even DDR SDRAM, which means that keeping these small caches filled is less of a problem.
Consider also the very long, 20-stage pipelines found in the P4. These also require very large memory bandwidth to keep them filled, and what memory technology does that better than RDRAM? Conversely, a branch-prediction miss by the P4 takes a long time to work its way through this 20-stage pipeline before the P4 can restart its work. This gives RDRAM breathing space to refill the P4's small caches from elsewhere in memory that it would not have if the P4's pipeline was shorter.
Cynics might argue that perhaps the P4's branch-prediction isn't as good as it could be, in order to make up for some of the access latency that RDRAM suffers from...
Don't forget the benchmarks either: where does the P4 really shine, and convincingly beat the Athlon? That'll be benchmarks where the data set that is being processed is large and (mostly) sequential. Why? Well, it's obvious, isn't it? It's down to RDRAM's massive bandwidth. Other tests which do not hammer the memory bandwidth as hard, or where the data set is scattered around memory, show the Athlon coming out on top.
All of this points to the fact that the P4 was at least designed with RDRAM in mind. If not specifically for RDRAM, then at least for a memory technology with large bandwidth and high latency, and how many types of memory do we know that fit that description?
I suspect that future revs of the P4, designed to work better with DDR SDRAM, will have larger on-chip caches, and (possibly) better branch prediction. The larger L1 and/or L2 caches will be necessary in order for the P4 to have competitive performance when using SDRAM. However, this will come at a price - a lot more transistors on the die. More transistors == larger die size, higher power requirements, and, worst of all for Intel, means that they will not be able to scale the clock speed up as fast as they would like to, which leaves me wondering whether they will genuinely be able to keep up with the Athlon in terms of performance.
I think among tech-types, this is a bigger embarassment to Intel than the Pentium bug back in the day. I mean, that bug was fairly harmless really, but it was blown up quite a bit in the media. Of course the actual seriousness of that bug is probably somewhere in between what Intel said and what the media said.
But for this Rambus fiasco, the tech community has scolded Intel pretty hard since day one with one benchmark after another, and Intel's refusal to accept that they may have selected an inferior product may have hurt them in the eyes of people who actually respect them for their technology.
Where are we going and why am I in this handbasket?
Brookdale was already in Intel's roadmaps for a long while. They're just waiting for their contract with Rambus to end, since it's preventing them from releasing DDR products.
Plus, they designed the P4 from the ground up to support RDRAM. It'd be just stupid to quit supporting them now, that it's becoming almost affordable to buy Rambus memory.
Personally, I loathe Rambus as much as the next guy, but if you think rationally, this decision should have been taken by Intel a long time ago -- now it doesn't make sense anymore.
Join the NFSNET. Our prime goal is making little numbers out of big ones. http://www.nfsnet.org/
I;m curious (and htis is NOT a flame - I'm really curious) If you've had the $$ for a new box and you want a speedster system - what kept you from getting an Athlon based system? The benchmarks I've seen between teh fastest P4s and Athlons haev been mixed - P4s excel in some areas while Athlons at much slower clock speeds excel in others the P4 can't seem to master. I'd paint it a draw unless all you do is play Quake :)
So is it brand loyalty? The need to have that ultra high (and meaningless) MHz number? P4s are more expensive, though sometimes the mboos are cheaper than equivalent Athlons.
I'm not trying to start a flame fest of replies - I'm just curious if maybe theres experiences out there that the numerous HW sites haven't touched on that make waiting for the i845 worth doing.
Yes, I tend to lean towards Athlons due to teh price/performance ratio. I can't stomach the prices Intel wants for some of their CPUs when a 1GHz Athlon 266FSB is now < $100 But in teh technical community that is /., besides brand bias/loyalty are there really black and white compelling reasons to shun Athlons in favor of a P4 with SDRAM support? Remember, some of the benchmarks that P4s excelled on were due to memory bandwidth of RDRAM and with SDRAM, they may not be so high - only time will tell of course.
So if you reply, try to be insightful instead of saying AMD sucks - who knows - this thread might be worth reading for folks who don't check the HW sites daily :)
Top Most Bizarre/Disturbing Error Messages
I have yet to see one and I don't know anyone who owns one. Im not convinced they even exist.
-And if the band you in starts playing different tunes......
my RAM won't cost as much as my damned CPU/Motherboard combo.
Screw 3...