See what i mean Gregz? Most people here hate X38 but they dont know why
PCIe2 is better to have than to not have and you can get good X38 boards for only $200, theres no reason to buy old tech when new tech is available for an added cost equaling to a few beers.
See what i mean Gregz? Most people here hate X38 but they dont know why
PCIe2 is better to have than to not have and you can get good X38 boards for only $200, theres no reason to buy old tech when new tech is available for an added cost equaling to a few beers.
I doubt anyone here "hates" X38, but P35 is obviously preferred...Why? because it overclocks better and X38 offers no real world advantages.
FYI my personal experiences were the exact opposite of that post for my Mars. I was on a P5K originally, and was able to increase FSB about 40-45 over that with the Mars, and was also able to get a about another 25mhz on my ram as well.
Believe me, all things being equal I'd rather get a p35 based on what I've read. It's proven, reliable, etc.
But...as I was saying...for the 3rd time...I'm not going to make any upgrades or changes to this machine for 3 years, except for the graphics card.
If a card comes out in 2 years (my guess is sooner) that can saturate the PCIe 1.1 16x bus, then that's at least a year of new games I can't play optimally with this machine.
If the x38 can do everything p35 can, as well as keep my machine in business for at least a year longer than the p35, why wouldn't I go with x38?
Greg, I hate to say it but asking for a computer to last gaming-wise for 3 years is impossible. You could drop 10k on a system right now and it'll be considered mid-range in a year and a half.
That being said, I'd still go P35. We're just now saturating AGP 8x like everyone else has said. MAYBE the highest end 9x card will begin to approach saturation, I seriously doubt it seeing as PCIe 16x has been around almost 3 years and it's sitting at a cozy 50% in most applications. But who can say what 10xxx will be? Just go for the proven stuff, have reliable performance on the super high-end anyway, and save yourself a buck.
^^
ecco hit the nail on the head... its taken how long for pcie 8x to even get close to being saturated... even with PCIE 2.0 spec out now, the time frame to saturate 16x pcie is prolly going to be atleast another 2-3years. no one can really say for sure since all we can do is wait for video cards to be released and test them on the different pcie speeds. even the current ati 3000 series is being made on AGP 8x and people keep saying that is dead, long time dieing then...
@th3
no body hates X38 here, we have pointed out why not to go with it... its all based on what you plan to do with your rig. had the OP said he was going CF then X38/X48 is the only choice for that route. both P35/X38 have the same life span as any other board. though P35 is about to be replaced with P45 which from what i can gather will bring PCIE 2.0 to the table. not sure if the Op will wait for that but i last read intel plans to launch it mid feb, i think its a paper launch though...
Why is it unreliable? I'm not saying that it's correct, but it certainly makes me cautious. Where is a similar graph that represents the % of PCIe 1.1 saturation vs. graphics cards that proves what you guys are claiming?
Greg, I hate to say it but asking for a computer to last gaming-wise for 3 years is impossible. You could drop 10k on a system right now and it'll be considered mid-range in a year and a half.
I completely disagree. You'd be right if we were in 1988, or 1998, but in 2008 the vast majority of gaming performance is determined by the GPU, which I plan on upgrading at least twice on this build. If you don't agree with that I'd suggest you do some reading. Also, the vast majority of gamers can't afford to buy a new rig every 18 months. I'm one of them.
Just go for the proven stuff, have reliable performance on the super high-end anyway, and save yourself a buck.
I plan on doing exactly that, in truth I'll probably stick with my 7800GT another 3 months until prices come down. But that isn't the issue at all, the issue here is building a PC now that will in no way bottleneck the performance of a high-end graphics card released 3 years from now. And I do think that can be planned for and that it's important to do so in order to "save a buck" down the road.
even the current ati 3000 series is being made on AGP 8x and people keep saying that is dead, long time dieing then...
"Marketing name HD3850 PCS AGP 512MB GDDR3 256bit
Model name AG3850 512MD3-P
Memory interface 256 bit
Core/Mem Speed 668e / 828m
EAN Code 47125050 22321
Output DVI+DVI+HDTV"
It has half the memory of the 3850 x2 and only one GPU...the current nVidia cards with similar specs could also probably fit in the 8x AGP lane, so I don't know why you're bringing it up since it's very likely that the next gen cards will be either multi-core or multi GPU and have at least 1gig of VRAM. Here is an illustration of the problem for folks who didn't click the link above:
"As you know, on the board ATI Radeon HD 3870 X2 is two graphics processor. According to the information available at the reference motherboard ATI Radeon HD 3870 X2 will be installed 1 GB of memory GDDR3, which operates at a frequency of 2 GHz processors and related 256 - bit tyres. The product is meant for connecting to the bus PCI Express 2.0, which interact with the responsibility of the special chip-bridge. The truth seems to be in this role will be used PEX6347 not, as expected, but PEX8548. At the given scheme, the bridge is at the centre of charges between the two processors. Near each processor can be found on four chip memory.
The following illustration shows that the transition from PCI Express 1.1 for PCI Express 2.0 provides a noticeable increase in productivity - around 20-30%, depending on the application and the screen resolution."
Yes that data is preliminary, but at least it's data. It clearly shows the 3850 x2 being bottlenecked by a PCIe 1.1 16x bus. If someone has a link to a review that disputes these results, I'm all ears.
greg you missed the point then of why i brought that up about the AGP geting a 3000 series GPU for it. the point was that AGP is dead and AGP 8x isnt being saturated. that goes to the PCIE 8x as well since both have the same speed, though AGP is 1-way not 2-way. my point also being that it will being that we dont know at what point pcie 16x will become saturated.
About with PCIE 1.0/1.1 vs PCIE 2.0 has been pre-NDA. we have no after-NDA numbers, none that i have bean able to find. on top of that comparing the 2 is highly wrong in the sense that both specs are on different chipsets. p35 -> x38, the only chipsets that would even be close would be 680i and 780i, since 780i is a tweaked 680i for 45nm quad support. the NB doesnt nativly support PCIE 2.0, thats brought in with a addin chip on the board below the NB. simply we cant go by that chart cause it shows a % increase. there are no hard numbers in that chart or where it came from www.pczilla.net/post/89.html
That is why "we" as a group cannot trust those numbers, they are %'s. where are the fps numbers? in WOC did they first get 30FPS? then a 25% increase making it 37FPS? you could get that same increase from ocing the GPU alone... who's to say they didnt do that in the first place? i find it hard to go with their numbers when they have no fps numbers for the setup. i may not be able to read whats on the page. i see no where on that page where any specs are listed for the computer other then the GPU specs.
we have done the best we can to help you with the rig you want to build. i get the feeling the only thing you want to do is argue something that you just cant. if we want we could play with photoshop to make fake numbers. there are to many other problems with tring to compare PCIE 1.0 vs PCIE 2.0. A card supportes PCIE 2.0 does that mean if i stick it in a PCIE 1.0/1.1 slot im going to see the same gains? what if i take a PCIE 1.0/1.1 card and stick it in a PCIE 2.0 slot, will i see the same? thats basicly what the source looks like to me... we have way to many unknowns that are not answered about that comparsion.
I think we can agree that most people don't have enough facts to answer this issue competently yet. I'm hoping that someone over the next week or two will come here and provide a link to a 2nd opinion review on this topic from a more trusted source (anandtech, etc.). A lot of people could benefit from that information.
Until then, I can wait a few more weeks before making a decision that will affect me for years to come.
Greg, you asked for advice. Many people have given it. These are all of these individuals’ unbiased opinions.
If you have a set plan, and don’t want any advice that’s fine. Many people already know what they want and don’t want to hear that others might recommend different hardware. No biggie.
But I’m a tad confused on why you asked for advice and now are not only dismissing it, but outright disagreeing with it and in a somewhat argumentative fashion.
I had a set plan to buy a P35 motherboard two weeks ago. After reading the responses and information in this thread (and other threads on other forums), and spending time researching the various issues, I have had to put that plan on hold.
So the advice has been very helpful, and it is appreciated. It's one reason why I (and thousands of others) come to forums like these.
I completely disagree. You'd be right if we were in 1988, or 1998, but in 2008 the vast majority of gaming performance is determined by the GPU, which I plan on upgrading at least twice on this build. If you don't agree with that I'd suggest you do some reading. Also, the vast majority of gamers can't afford to buy a new rig every 18 months. I'm one of them.
I can counter your point by pointing out in the last three years we've gone from single-core processors at 130nm to quad cores on 45nm. When I put together my AMD 4000+ system with 2gb a ram and a 6600GT I was convinced mine would last forever. However a year and a half later (this time last year) it not only showed it's age but was considered basically a budget base-line gaming system.
Yes, quad and dual cores have changed the way processing works. But with IBM having a working light-based 500Ghz-equivalent, who's to say what changes will occur in the meantime.
I"m on your side on this on, I can only afford to build a new computer every 2 years or so, with a processor upgrade usually every year (due to my employment accomodations.) And believe it or not, I'm subscribed to see what mobo you end up with as my current build lacks a good mobo and my old graphics card needs to be replaced (see my sig)
As for my advice, spend moderately, don't jump on the best of the best. And stash the cash for a mid-build upgrade sometime down the road, it's what I do
Hi phu, thanks for responding. Unfortunately that link doesn't address the problem we're discussing.
From the article:
To get right to the point, upgrading from PCI Express 1.x to 2.0 is not worthwhile right now. The current crop of graphics cards just doesn't tax the PCI Express bus enough for a difference to be visible.
Noone is arguing that the current cards are taxing the PCIe 1.1 bus. The problem is how to build a gaming machine today that will allow you to keep a relatively current graphics card in your machine 2 or 3 years down the line.
It remains to be seen what the situation will be when newer graphics cards such as the Geforce 9 or the Radeon 4xx0 begin transferring larger amounts of data over the bus. At any rate, the P35 and 975 chipsets are easily sufficient for the current generation of cards.
So basically they're saying, if you plan on building a gaming machine now, and don't plan on ever upgrading your graphics card, then P35 is fine for you. Most gamers, including myself, plan on upgrading their graphics cards at least once over the lifetime of their build.
Anyway it should be relevant that the speed of 1.0 vs 2.0 is the same. As they can't review any future gfx-cards to get you the answer you're looking for, if that's not the answer you're looking for. I wonder what it is you're really looking for?
I don't see the problem where the currect PCI-E 1.0 x16 shouldn't be sufficent for newer GPU's in the future ? Since it's not slower than 2.0 x16.
Cause you where discussing about p35/x38 PCI-E 1.0 vs 2.0 right?
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
