Mid~Late November, in time for X-mas.
Printable View
Mid~Late November, in time for X-mas.
I like TPU but there is something wrong with this review. Don´t know about the results of the HD5870 but when I look the numbers of other cards like the HD4830 and HD4850 being just exactly the same in all tests I have to say that can´t be ok.
so, say I have a 4870 1gb and a 4870x2 together in trifire with a FC block on the x2 and a mcw60 on the 4870... Would it be worth it to switch to 5870CF and could I keep my mcw60 or should I get different blocks or just not switch? (for gaming at 2560x1600 on a q9650@4.2ghz).
oh :banana::banana::banana::banana: you are right. i must have used the wrong data for one of those cards ... i'll check and remake the graphs tomorrow. thanks!
edit: both raw benchmark data files indeed are the same .. since this can't be right i'll have to rebench the 4830 or 4850 tomorrow as well :(
this has been wrong for months, and you are the first to point it out ... congrats
given the price of a 5870x2 vs monitors you may reconsider that soon. i just saw a sale for 23" 1920x1200 (or maybe 1080) for 160$, give it a year and the no bezel monitors could be probably under 150$ each, and 3 of them i think would look so much better than adding a second 5870 on 1680x1050. and at that resolution (which may be considered low soon) will probably be cpu limited way too quickly. to summarize, 1 5870 with a fast cpu i think will be more than enough for any dx11 game in the few years.
There is something not yet address in full that's also been speculated. That is the use of 5870 and Win7 vs Vista. It's been rumored that 5000 series would provide better performance under Win7 as drivers mature. However, that remains to be seen. Also, there are a lot of mixed results when trying to assertain performance when so many reviewers are either using:
-Win7 32
-Win7 64
-Vista 32
-Vista 64
Specially when games are 32 bit centric. And, no measures were taken to tweak 32bit games in a 64bit OS so they can take advantage of the larger virtual address space that 64 bit OS can offer (until they note otherwise of problems using large address space flag, etc). However, it does appear to me that at it's current price point it safe to suggest the 5870 as we will see DX11 game(s) this year and next year.
If all of you take a look, AND read the 2 reviews, you will be able to notice that the results with the OLD drivers and the SLOWER E6750, are better than the others. I'm very disappointed because I've written them a letter and nobody even bothered to write me back. These reviews are more than 2 years old. How should I trust them?
my cards have the right shaders for sure, after all i found out http://www.techpowerup.com/articles/other/155
as i said before, we changed the settings. if i remember correctly we used a different shader model / lighting algorithm in the newer setup
Regarding caches - yes, but not only.
4 L2 caches stayed from RV770, capacity doubled. Each L2 is for 64bit of mem, bus. So that cache now can be managed more flexibly because it's twice as big. Similar thing to doubling shared L3 cache on CPU. Some code will execute faster because it will have more data available (like in case of single threaded task).Quote:
However with twice as many SIMDs, there comes a need to feed these additional SIMDs, and to do something with their products. To achieve this, the 4 L2 caches have been doubled from 64KB to 128KB. These large L2 caches can now feed data to L1 caches at 435GB/sec, up from 384GB/sec in RV770. Along with this the global data share has been quadrupled to 64KB. /Anandtech/
Very good point with interpolation moved to shader core. For some games this will be perf. improver and for other it might be slight bottleneck.
Has anyone seen the 2GB versions yet? Also, any whispers as to a 4GB (2x2GB) 5870X2? After a long stint w/ Nvidia, ATI has pulled me with the 3+ monitor support. Been waiting on that for /years/.
Mmm will they have the same performance on a x58 board , or will they run better on a AMD board ?
OOS now, but heres a diamond at ZZF:
http://www.zipzoomfly.com/jsp/Produc...dlist=celebros
Hmm... how to get more "memory space".
Recall pointers and segmented memory space on 286? F-up ugly hacks, but "works" - sometimes... ocassionally... alright - rarely.
I understand the reviews were done quickly for launch, but hope some of the reviewers will try:
- OC CPU to >4Ghz. ie 975 @ 4.2Ghz: half the games in reviews like L4D and Wolfenstein are 150-200fps at ALL resolutions up to 2560x1600 4xAA.
- compare 5870 CF on Bloomfield and Lynnfield (and sure why not, Phen2).
- run oldschool games like Far Cry at 999 fps LOL
- Once 5850 and 5870 with faster GDDR5 chips arrive, can try running at low and high GDDR5 speeds at same GPU clocks to see SP scaling and where bandwidth issues.
- alternatively, try clocking 4870x2 to 850/600 to mimick a 5870... results in HAWK, Resident Evil 5 and Batman, where 4870x2 is 22-42% faster likely due to bandwidth, should go down and be equal or lower than 5870... otherwise we got a problem ;)
I wanna see some comparisons between processors on these cards
I have a q9550 at 4 ghz... but I wanna see how much slower it is than some 4.2 ghz bloomfield... I mean like half the games were cpu bottlenecked with 3.8 ghz bloomfields up to 2560x1600
5870x2 looks DEFINITELY like my new buy... I want to see the new asus Voltage Tweak Tuner or w/e it is on a 5870 x2 with 4 gb vram (2gb effective)
would totally rock on wcing
Fun Facts:
* Xeon Lynnfield supports max 32GB DDR3. Since X58/P55 dont support ECC, they only support 16GB :<
http://en.wikipedia.org/wiki/Lynnfield_(microprocessor)
http://ark.intel.com/Product.aspx?id=42928
with 2GB cards like 4870x2, graphics memory is projecting to surpass desktop memory - maybe we can use for swap file or something l;0-
* 5870 like 4770 is made on 40nm. Intel and AMD CPUs have traditionally lead being first down to a smaller lithography. Only other time that happened was 55nm RV670 - aka HD3870 - which beat 45nm Wolfdale/Penryn by a couple months in winter '07.
* Not sure, but 5870 might be first "consumer" chip with 2Billion + transistors. To put into perspective, Intel 6 core 500mm^2 Dunnington with hefty 16MB cache only has 1.9B transistors.
* Also not sure, but this is 5th GDDR5 product, and all were from AMD/ATI.
* Remember 7900GTX @ 650Mhz. Only other nVidia GPU with higher clockspeed is 9800GTX/+ (675/738). nVidia seems stuck around 650 :(
* Meanwhile, many AMD GPUs launched >700Mhz (even 80nm 2900XT). But, only 3 I know higher than 800Mhz. 850Mhz HD5870, 850Mhz HD4890, and 825Mhz 3870x2 *shock*
* AMD's first GPU with 32 ROP. nVidia already made transition to 80 tex, 32 ROP in 1.4B transistor 65nm GTX280 a year ago. Coincidentally, when GTX280/GTX285 launched, low idle power was also a highlight.
Just found 2 comparing i7 with PII
http://mikrodatorn.idg.se/2.1030/1.2...ger-dig-svaret
http://bjorn3d.com/read.php?cID=1682&pageID=7612
Fascinating... amd system performed better in some cases... must be the improved clock speed
I may have missed it. Anyone get some photos of the naked board yet?
Nope, I like my monitor a lot even have HDMI on it been waiting for new cards to finally use it. I won't run multiple monitors either think it looks stupid having pieces of plastic between screens. Be okay for flight sims but that isn't really my bag. Whena re 5850s coming out?