Saturday, May 31st 2008
![ATI Radeon Graphics](https://tpucdn.com/images/news/ati-v1723816949774.png)
AMD Starts Shipping ATI Radeon HD 4850 Video Cards
The ATI Radeon HD 4850 cards are reportedly already shipping to OEM partners and retailers. According to TG Daily everything is going as planned and AMD/ATI is aiming for a sizable launch of the new product generation, with Radeon 4850 512MB boards leading the charge. The final prices for all Radeon HD 4 series cards will be officially declared during the Computex 2008 tradeshow which will open doors on Monday. Higher-end Radeon 4870 cards with 512MB of onboard GDDR5 memory are expected to ship in volume sometime this summer, with flagship Radeon HD 4870 X2 to follow soon after that.
Source:
TG Daily
74 Comments on AMD Starts Shipping ATI Radeon HD 4850 Video Cards
im hoping this will help AMD regain some of its popularity with its gamer fans.
HD3000 series was nice. My 3870 is serving me well. Upgraded to it from my x1800xt and haven't regretted it. It would be nice if you could Xfire this with the 3000 series, but my board only has 1 PCI-e slot :shadedshu .
It has already been proven that the memory bus in particular isn't the greatest factor. ATi already tried the 512-bit memory bus and it didn't last well. nVIDIA had 320-bit and 384-bit and notice that the G92 core with 256-bit is better performing in AA/AF benchmarks?
When you can have a clock speed of 1.125GHz on your memory it's massive bandwidth even with a 256-bit bus. Not to mention the additional benefits and performance improvements of the technology.
And if you never noticed, nothing above GDDR3 has shown significant gains on ATI cards, even when all else is equal. So the bandwidth of both GDDR5 on a 256b bus, or GDDR3 on a 512b bus is essentially overkill.
GDDR5 isn't unnecessary, it's unattainable by nVIDIA because they aren't involved in the development of it. Come to think of it, didn't ATi help develop GDDR3 as well? The same memory technology that nVIDIA is using on it's flagship cards?
en.wikipedia.org/wiki/GDDR3
Yes it is! ATi and JEDEC developed it! I guess some companies are still followers on the market, eh? ;)
. . . come to think of it, we haven't really seen anything brilliant or noteworthy come out of Intel since they slapped two P4's together on the same die. IIRC, even the looming-sometime-this-decade-release Nehalem is still milking the core2 architecture. sad, too, ATI jumps on new technology as soon as they can, and nVidia lags behind - they remind me of Intel in many respects, as long as their current tech works and dominates the market, why change it? If they get ahead of the competition, they sandbag their current tech and milk it for as long as they can.
ATI was the first to offer GDDR3, GDDR4, GDDR5, DVI, HDMI, PCI-E 2.0, DX10.1 (they also would've had the first DX10 release, if the merger with AMD didn't hold up the HD2000 series as long as it did), etc, etc - and people will purchase their products simply for the new tech, but it doesn't make up much for their lagging performance :(
but, y'know, IMO, if AMD designed GPU cores as massive as nVidia has, we'd have seen a different story the last few series.
Currently quad core processors are out, soon we'll see 8 core processors, etc. It's still advancing and heavily. The real thing is applications making use of SMP so they gain from these multicore processors.
2- Right now slow GDDR5 is 10-20% more expensive than higher speed GDDR3 (0.8 ns). Slower GDDR3 such as 1ns memory is way cheaper. GTX260, the only possible "direct" competitor to Ati, uses the cheaper one and has more or less the same bandwidth as the 770XT. GTX 280 has a significant higher bandwidth, so we are not comparing apples to apples there. We could easily say that we are comparing extreme high GDDR3 vs. slow GDDR5 and the actual result is Nvidia has the higher bandwidth, so Ati solution being cheaper makes sense, it's what we should expect from the slower part. 512 bit + GDDR3 here is giving more bandwidth, it's overkill, unnecerary IMO, but faster still.
3- I didn't talk about the implementation, but the price of memory. If you read his post you would notice he first says GDDR5 is cheaper and then starts talking about the implementation like this: cheaper GDDR5 + cheaper controler, PCB, etc = WIN WIN. That's what I understand in his statement. And that's not true NOW, it will in 3-6 months probably. And that's what I stated. And it's in this same moment, having to justify my statements, when newtekie's post #20 makes even more sense to me... Indeed! Who said the oposite? I only made clear that GDDR5 is not cheaper NOW. I didn't say it was a bad decision or anything. I just say that price wise NOW may not be better, not by much at least. In reality we don't know if it's cheaper at all:
1- Memory is more expensive and in short supply NOW*. So much that Ati halved the frame buffer to 512 MB in the 770XT and uses GDDR3 in the Pro. Never forget this, as it pictures the truth in a "check the reality" fashion.
2- We have no real proofs that making a 256 bit GDDR5 controler is cheaper than 512/448 bit GDDR3 controler RIGHT NOW* and get the same performance. Remember DDR vs DDR2, DDR2 vs DDR3, GDDR3 vs GDDR4...
3- PCB is a lot cheaper, but IMO it's the only part where there's an econ0mic benefit RIGHT NOW* and it doesn't account all that much to the retail price..
*Sorry because I used it so much, but I think that I have to make clear that I am talking about today. Otherwise some people will think I'm saying something that I am not. I have no doubts it will be better with the time. That's why I said 3-6 months.
That's the sadest thing I have heard in a long time. It's extrememly unfair and bad for free competition. :shadedshu
I had no clue Ati was involved in the development of GDDR. I thought it was the JEDEC who does this things in conjunction with memory developers in any case (when talking about memory of course). And not only ONE of the consumers. No wonder why Nvidia is not using it! Ati has probably tons of patents that doesn't want to share for cheap! If they want to share them at all. :banghead:
Not to mention that this way Ati is kind of imposing the use of the memory the way they want it to be, which may not be the better way, who knows? In any case it surely benefits Ati. It's not like other companies can't develop their own standard, or that they can't make it better, it's surely more based on relationship with JEDEC. It's like TWIMTBP in hardware development, with the exception that there's no patent involvement in TWIMTBP and in harware it's SURE there is. :shadedshu
And implementing technology for technology's sake is hardly the right way to make money. Did GDDR4 prevent the 2900 XT from being a POS? How is DirectX 10.1 useful if not ONE game in existence uses it? In this industry, performance is king - ATI can provide all the features they want, but if they don't provide the horsepower to go with them, no-one will buy their cards and hence no-one will use those features. By that logic, all ATI has to do to regain the performance crown is release a GPU with 2 billion transistors.
As for the GDDR3 vs GDDR5 debate, it's been done to death already. The fact of the matter is that nVidia would be suicidal to go with GDDR5 because there's so little of it available, it's hideously expensive, and (almost certainly) because ATI has already purchased most of the stock that's available.