What you don't seem to understand is....if I was OK with the problems inherent with multi-GPU, I would've done so for much less money months ago. The fact is, deciding not to go SLI/CF isn't something that just dawned on me when the 4850 came out, I've been weighing the pros and cons for almost 2 years. The reason price does not bother me is because there was simply nothing I could buy as a single-GPU solution that offered anything close to GTX 280 performance until GTX 280 was actually released. That's why price didn't matter, and still doesn't matter. Its really that simple.
What you don't seem to understand is....if I was OK with the problems inherent with multi-GPU, I would've done so for much less money months ago. The fact is, deciding not to go SLI/CF isn't something that just dawned on me when the 4850 came out, I've been weighing the pros and cons for almost 2 years. The reason price does not bother me is because there was simply nothing I could buy as a single-GPU solution that offered anything close to GTX 280 performance until GTX 280 was actually released. That's why price didn't matter, and still doesn't matter. Its really that simple.
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Really (at the bolded words)? link? i know that the PLX chip on the R700 does help against the micro-stuttering issue but have not heard about a "fix" for dual/multi card setups since there is no bridge chip involved.
Keysplayr
Elite Member
- Jan 16, 2003
- 21,209
- 50
- 91
Actually, I believe Chizow is getting some serious cash back. So now it's no longer twice the price? And you use the term "inferior performance" too loosely. Travel around some more review sites and you'll see various differences. GTX280 trades blows with Xfired 4850's.
And when Xfire doesn't work, well, you know what happens then. The money Chizow end's up spending is a little more "solid" for a lack of a better term. He can count on consistent gaming performance for pretty much any game. You probably don't have that luxury.
In the end though, are you both happy with your purchases? Because if you are, that's all that matters, right? No need for the swagger from anyone. Yes, that includes both you and Chizow. Your conversation is going nowhere.
Now, about that GT200b?
Who thinks there will be other improvements to the core with the die shrink? Will the need for the NVIO chip remain, or go away like it did going from G80 to G92?
Do you think it will still be 512bit?
You're right on, pros and cons to both. But $380 more for a single GTX 280? Uh...ok, if you say so.
Uh, put your glasses on and read the review: http://www.guru3d.com/article/...ossfirex-performance/1
Then come back and repost "the 280 absolute owns in 2560x1600"
evolucion8
Platinum Member
- Jun 17, 2005
- 2,867
- 3
- 81
Nobody had complained currently of having Micro stuttering issues with HD 4XXX in CF, so find me a link that states that there's Micro Stuttering on the HD 4XXX in Crossfire, it was something easy to pin point when the HD 3870X2 debuted or when people did Crossfire with the HD3XXX series.
Wreckage
Banned
- Jul 1, 2005
- 5,529
- 0
- 0
That is my experience as well, no micro-stuttering observed at all. I have only replied to this thread to quell the mis-info. Funny how most all of the 48xx CF reviews make no mention of micro-stuttering either. This is what happens when people that are not qualified to make comments comment anyway :roll:
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
In the tests where they actually compared the GTX 280 to 4850CF and 4870CF, the GTX won some and lost some. But they did not provide the MIN frame rates, which I have seen elsewhere, in which the GTX 280 does absolutely own at that resolution.
Also their review is erratic, sometimes they use the GTX260, sometimes the 280... sometimes they compare to a 4870x2 + 4870 in CFX to a single GTX280... weird review
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Im sure you read the thread on micro-stuttering issues yourself and should know by now that all multi GPU setups (CF and SLi) that use the AFR method suffers from it. What im saying is that micro-stuttering does exist and ive yet to see a decent fix for it either than the custom PLX chip used by R700 that relieves this problem somewhat.
These hardware problems do exist and are probably documented as a errata in a long list of other erratas. They dont just magically disappear either when no ones complaining about them.
The problem is that the multi-GPU market makes up a very very small % of the entire desktop market. Its always been a niche product (was very expensive) til now. So when there is a problem or issues it might often get dismissed or most users may not be familiar to what they are talking about. Compare this to the "shimmering" problem. This was well documented issue and people knew what it was exactly because alot of users had G7x/NV4x type hardware that could test/investigate on the issue.
Some people observe no micro-stuttering. Thats great for them. But for some people, they get irritated by it because they can observe it. There is no mis-info. Why isnt there any word about it in the reviews? because micro-stuttering isn't observed at all when running benchmarks (timedemos, etc etc) nor does it have any effect on the fps result. Its something a user experiences when playing a game, not when he/she runs a custom timedemo then decides to go have a cup of tea.
Not qualified? let me ask you a question. Just because you own a 4850CF setup means that you know how it works? that you know how AFR works? well since your "qualified" according to you, then i guess a whole host of people are wrong about this subject including major IHVs that are wasting their time with shared memory pool technology and other technologies that are aimed to tackle this issue to provide a more flawless multi GPU solution.
Keysplayr
Elite Member
- Jan 16, 2003
- 21,209
- 50
- 91
Heh, page two of your Guru review.
Uh oh ! With multi-GPU setups, you are bound to run into problems as well. We surely had them as for example Frontlines would not even start.
What are you qualified to make comments on btw? And what is it exactly that "qualifies" you?
Qualified to make comments on what I use, contrary to a whole slug of people who don't own jack squat. I didn't say CF was the end all solution, and wow, a few games have probs, is that really a suprise? I take issues with that lame bullet list Chiz made (which applies more to old CF) and the fact that he doesn't even own the CF product. Same for you Keys, what qualifies you?
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
I devised methods for testing for it. I was getting micro stuter on a single GPU (8800GTS 512 and 4850) in mass effect on an E8400 OC. I tracked it down to CPU speed, and I found a way to document and test microstutter.
I used fraps to measure exact frame times, then I created another column with the formule to get the exact time to render each frame.
I copied the data only, sorted them, and then searched for exactly the same numbers shown (like, 41.32ms to render a frame)...
I found that on occasion those high ms to render a frame were preceded and succeeded by low ms. 41ms = 23FPS.
An example was
Frame 661: 10ms
frame 662: 41ms
frame 663: 15ms
frame 662 is micro stutter in action. it drops from 100fps to 23fps and then goes up to 66fps.
This prompted me to upgrade my CPU from an E8400 @3.6ghz OC, to Q6600 @3.0ghz oc with undervolt. (I got higher OCs with over volting, but come on, huge OC with undervolt? that is just awesome!)
They are all making comments based on some reviews that point out scaling problems...just like you are pointing to reviews that show 4850CF beats the GTX280.
This thread has become rather cyclic hasn't it?
This thread has become rather cyclic hasn't it?
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Has that helped?
Man we are going way off topic. To get back on track, i think nVIDIA might not just go for a die-shrink to 55nm but actually cut off the unnecessary fat (the GPGPU related i.e FP64 support for example) to reduce transistor count and die size. I think that nVIDIA was trying to "kill two birds with one stone" i.e have a great gaming solution and a GPGPU solution.
Currently, i think its better be a master at one thing than being the jack of all trades since that produced a 576mm^2 monster. So in the end, i think nVIDIA should immediately get rid of the GTx260/280 and prepare for the GTX270/290 asap i.e the 55nm G200b.
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
I am still tweaking the CPU OC (I installed it last night, I OCed it today), I will do benchmarks for mass effect later, and will use thread affinity to limit it to 3 and 2 cores as well to see the difference.
I agree that they should trim the fat, but that might be too time intensive. They already transitioned SOME of their line to 55nm and they stated that the G200b (55nm G200 chips) based GTX280 will be released on august 25th, which means 31 days from TODAY. I hope they don't mean paper lunch, but none theless this was QUICK!
They already announced that the succesor chip will have DX10.1, and will use a smaller bus with GDDR5 chips. and a few other things I forgot.
I agree that they should trim the fat, but that might be too time intensive. They already transitioned SOME of their line to 55nm and they stated that the G200b (55nm G200 chips) based GTX280 will be released on august 25th, which means 31 days from TODAY. I hope they don't mean paper lunch, but none theless this was QUICK!
They already announced that the succesor chip will have DX10.1, and will use a smaller bus with GDDR5 chips. and a few other things I forgot.
evolucion8
Platinum Member
- Jun 17, 2005
- 2,867
- 3
- 81
All the fan boys love to jump against to only one person like if they were paid by someone. If he says that he doesn't have microstutter with his setup, who can prove him wrong? That doesnt happen always, and now is even harder to happen, or will you guys buy a Crossfire setup or just go and visit him to prove him wrong? nah, so just keep going to the thread or get a life, being all day in front of the computer is ain't healthy.
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Evolucion8, if you want to say something useful or insightful at least read what we were discussing about or post about what this thread is all about instead of making rather bold statements to other AT users and then go as far as to being very rude dont ya think?
Damn that is...almost dirt cheap! Show us the 300 mark, make us happy!
Just for the sake of comparison, a leadtek gtx 280 in Denmark is labeled ''dirt cheap!'' at a cost of 620 bucks. Yup, ain't that cheap...
Keysplayr
Elite Member
- Jan 16, 2003
- 21,209
- 50
- 91
Anyway..... If the current GT280, for example, can o/c to 700/1500 on a 65nm process, I think we can look forward to some nice stock clocks on the 55nm GT200b's. And Nvidia should be very competitive on the price point. I do hope they keep the 512bit bus however.
ViRGE
Elite Member, Moderator Emeritus
- Oct 9, 1999
- 31,516
- 167
- 106
Today's forecast: Cooler heads will prevail, or this thread will be locked by sundown.
I don't think I need to yell at each and every one of you, you know who you are. Stop posting purposely inflammatory comments or you'll be enjoying the great outdoors (because you sure won't be here).
I don't think I need to yell at each and every one of you, you know who you are. Stop posting purposely inflammatory comments or you'll be enjoying the great outdoors (because you sure won't be here).
Once again, just because you guys don't "hear the music" or notice the phenomena, for whatever reason, does not mean microstutter doesn't exist. I've already challenged someone else claiming the same to prove it doesn't exist rather simply, all it involves is using FRAPs to log FPS and post the results so that we can take a look at them. Ideally you use a game/sequence that varies in FPS and drops below refresh rate, as frame rates higher than refresh will normalize delay and mask any microstutter.
512-bit and GDDR5 ideally.
But ya my concern about changing bus width and memory controllers would be any impact on ROPs. I remember seeing a labeled die shot though and it looks like they rearranged the layout somewhat so that the ROPs are centered on the die at the top and bottom instead of spaced around the edges like with G80. Hopefully this means they can cut controllers without also cutting ROPs as was the case with G80/G92. The other big improvement to bring it in line with G92 would be shader clocks. There will obviously be some core clock improvements with 55nm but GT200 really lags with shader clocks compared to G92. Its actually pretty amazing that a 9800GTX+ actually has higher shader op/s and texture fillrate than a GTX 260 due to clock speeds. GT200 gets some improvements with FP64 but overall GT200 shader op/s isn't anywhere close to 2x a G92 running a shader core clock of 2000MHz because of the clock speed difference.
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 10K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
-
-
Facebook
Twitter