Discussion in 'AMD / ATI' started by LAN_deRf_HA, Dec 22, 2011.
RINGA-DING-DING we have a WINNAR!!!
We really don't know what method is used to allow PCIe 3 to become PCIe 2 compliant do to arch difference, bios which can be updated, etc. It's better to use a different MB that supports PCIe 2.0 in this example. Have a look at KitGuru results using a x58 MB. Using 2 different mbs are important here when testing PCIe performance. Sure you can add information about what the x79 would show at PCIe 2.0 but IMO, using a separate MB (a SB MB) is important in this case.
One shouldn't do it with an I7 when you want to know dual to core or dual to quad performance. Also one shouldn't do it when you have a 7970 and want to know the results of a 5870 and so on.
Here is an interesting post on performance % between the 7970 vs 580 and 480 vs 580.
Also there is news (be it rumor or otherwise) that Kepler maybe (at least) 6 months away
This is the only hint I could find of 3.0 support on SB-E CPUs, seems vague though.
I understand your qualms with not knowing how PCIe is implemented, but you msut understand that using two different motherboards is the most foolish thing ever, and is quite suspect, as the perforamnce compare can in no way be validated due to the myriad of timings and changes in BIOS that end users have no access to.
From having now reviewed some ~20 motherboards in the past year, it's become painfully obvious to me how important a board's BIOS is. Performance and power consumed can vary from board sample to board sample, never mind from two different products from differing product lines, then you want to add in BIOS differences?
The only way I see this as being feasible from the standpoint you prefer is to use one of the many 1155 boards that have been updated to "support" PCIe 3.0. technically, there is very few physical changes required, and this would allow use of hardware that is as close as possible to being the smae. but gain, the BIOSes will be different, so no true direct compare is available, and your method, or my method, is equally viable.
Your post is without merit and you've not provided otherwise to substantiate it other then to get mad about it . I find no reason to not allow the same test methodology that KitGuru used. Per their own results there is a noticeable difference between 2.0 and 3.0. So your suggestion that using just x79 is at best whimsical (based on your response) if for nothing else to start an argument.
So I hope to see more results using more then 1 motherboard. It will provide further insight on what the 7970 can do with the various motherboards that people are currently using. It's unfortunate that you don't want people to get an idea of what to expect based on their on chipset. What you are implying is that if someone has a P67 (for example) and want to get an idea of what the PCIe 2.0 does against x79 using PCIe 3.0 you suggest that they only use a x79 mb with the dual PCIe 2.0/3.0 feature. Simply because a x79 board offers the feature.
No, not at all. I think you need to compare 2.0 vs 3.0 on the same chipset, and it should be as simple as using the same board, and the right CPU.
For example, skt2011, use SB-E for 2.0, and IB-E for 3.0. All other parts identical.
For 1155, use SB for 2.0m and IB for 3.0. All other parts identical.
It's literally that simple. Of course, if the IPC of the IB/IB-E CPUs is better than SB/SB-E, that's suspect too. Otherwise, you need PCIe 3.0 CPU, and a PCIe 2.0 CPU...and two boards..one with PCIe 3.0 hardware, and one without.
My point is that no matter the testing method, there are caveats that must be considered. There's not really any direct way comparing the difference. I don't care what KitGuru did...if they did post results with 2.0 and 3.0, using current CPUs, they made a mistake, as 3.0 isn't supported, performance difference or not. SB-E does not officially support PCIe 3.0.
Truth is, you don't know what allows x79 to use PCIe 2.0. Furthermore, people would like to get some idea of how their PC compares to x79 setup. Just using a x79 does not provide that information and really have no way of knowing how it will relate to our setup. Because we don't have x79/3960 setup to make any comparisons based on what they currently have.
So in the end, using more then one motherboard is needed.
besides in the end it dosent matter because the PCIE 3.0 performance right now is worthless and why you ask? because no end user on earth right now unless they have access to industry insiders has PCIE 3,0 working if i had SB-E rig i wouldnt be able to get a Bios that supports PCIE 3.0 give me a platform that natively supports it and test that way, aka use Ivy Bridge on say Z77 and Z68 for 2.0 and 3.0 thus cpu and ram performance is roughly same, only difference is mobo which would be 1% margin of error,
but thats jsut me right now theres no way for an end user to test PCIE 3.0 and as such no official support means its suspect a performance increase is always nice but right now PCIE 2.0 #s are the only numbers that can be compared across the board for a good idea of whats going on.
I agree which is what I'm trying to get across to him, . There is no reason to only use a x78 setup to get 2.0 and 3.0 results.
main problem is theres no Ivy Bridge and NO Z77 yet and SB-E is it and its not officially supported thus suspect, which is why i agree with dave that right now its all worthless, i could make an educated guess on its increase and post fake numbers it dosent make a difference If PCIE 3.0 does change things the false numbers can be blamed on premature BIOS, this is why right now its worthless info.
I agree with you cadaveca, to get a proper representation of Gen 2.0 vs Gen 3.0 values a reviewer, or whoever wants to test it, should ELIMINATE all the variables that could give you different results thus using the same setup and switching ONLY CPUs
Um, first, let's get this right, please. It's X79. Not X78. There is no such thing as Intel X78.
And yes, I do know the difference between PCIe 2.0 and 3.0. It's about the encoding method used(8b/10b vs scrambling), and the speed of the interconnect. PCIe 2.0 runs @ 5 GHz, and PCIe 3.0 @ 8.0 GHz. X79 itself doesn't use PCIe 3.0, at all, BTW. Just the IB/IB-E CPU support PCIe 3.0. The Intel X79 Express PCH is a PCIe 2.0 device. I think YOU are not sure what the differences are.
Also, Intel states that 2.5 GT and 5.0 GT are supported, and the SB-E processor is CAPABLE of UP TO 8 GT.
Again, board is not the problem...CPU is. To properly test the difference, you use the same board for the platforms that offer it, 1155 get a board, and 2011 gets a board. You change CPUs, enabling real PCIe 3.0. Comapring the difference between differnt CPUs will only tell you the difference of the PLATFORMS, not the PCIe protocol used. You are missing my point. You don't need to test current/past platforms...the countless reviews published today already give you those numbers. the only way to properly document the difference in 2.0 nd 3.0 is to eliminate all other variables, period.
Ok, I said x78 instead of x79, got it now . Truth is, you don't know the details involved in what allows the x79 setup to use PCIe 2.0. Sure you can provide links about the benefits of PCIe 3.0 but nothing about how PCIe 2.0 works from a PCIe 3.0 arch.
The only thing you will show using only x79 setup using PCIe 2.0 and 3.0 will be just that. x79 results using PCIe 3.0 vs PCIe 2.0. Nothing more nothing less. This is not what I'm talking about. I want to see x79, PCIe 3.0 performance with other PC setups like i7, etc. And yes, that can be done. Because if provides an insight as to what the different PC PCIe setups offer when the 7970 is used.
I am impressed with the idle power consumption.
Again, the chipset isn't involved. All graphics ports directly connect to the CPU.
And agian, the difference is the encoding protocol, and the connection speed. Clearly YOU do not understand.
As to comparing PCIe 3.0 platforms to current platform, yes, I understand why you'd like this. However, because PCIe-supporting CPUs, which the videocards connect to, won't be release until the spring, you are left waiting until then for that. Any current numbers given are not done with parts that officially support the higher interconnect speed, or encoding methods. The actual electrical differences are available on the PCI-SIG site, if you are really that interested.
Please do not forget, I responded to this comment of yours:
To which I stated you do not need two different boards to compare the difference, becuase the board is NOT just what determines whether PCIe is supported. Please keep in mind all of my comments are based on that comment of yours, not what you've now changed the discussion to. Nice try, though.
It's been said that some of the review cards had the wrong thermal paste on it. I provided a link about it a page or so ago before the PCIe debate . Using good thermal paste did reduce power consumption, fan speed thus noise and temps.
But yeah it's impressive. I wonder if it would make any difference if we still replaced the thermal paste enough they have the correct one from AMD?
Nice strawman argument. I never made mention regarding any relationship between PCIe and how it is involved with chipset itself. I only use the p67/z67/x58,etc to make reference to that PC setup. Also, I've not changed the subject that we need more then one mb to see PCIe 2.0 vs PCIe 3.0 results. And to conclude, you provide no answer as to how x79 PC setup uses PCIe 2.0 using a PCIe 3.0 arch.
In any case, you are going around in circles about this. And I stand by my statement that we need more then one motherboard to see what a 7970 does with P67/Z67/x58 vs x79 (setups) to see the PCIe performance. If all you want to see is the x79 PCIe setup performance then that's something entirely different from what I'm talking about. And by your reply that's all you are going to get out of it.
Again, X79 doesn't support PCIe 3.0. It supports PCIe 2.0. You are clearly not understanding how this all works. I'm more than willing to explain it again...and again...
Again, the current SKT2011 SB-E CPUs are PCIe 2.0, not 3.0.
Official diagram from Intel, which I obtained from Intel for my reviews. Note no mention of PCIe 3.0.:
And yes, I'm going around in circles. WEEEEEEEE!!!!
Wow, you don't take well to being told something do you? But since you want to be specious about x79 here is the link to Asus's Rampage IV Extreme
As you can see it offers both PCIe 2.0 and PCIe 3.0 and...wait for it...it's an x79 board .
But in any case, it's been a good laugh. I look forward to more reviews . But hey, at least you admit to going around in circles when you really had nothing to rebut my post(s), weeeeeee!
Yes, the board supports PCIe 3.0 and PCIe 2.0. That is thanks to these(my own pic from a different board. The ROG board uses the same part.):
But that is not all....there are s a fe other board-level components required.
What is the real deciding factor between PCIe 3.0, and PCIe 2.0?
The installed CPU. And because the current CPUs don't support 3.0, but only 2.0, all that these BIOSes are doing is moving these ASMedia Or Pericom switches into the right mode, but perhaps, not the CPU.
Weren't you both agreeing that the problem was the CPU? lol
Sure, then ECH went on a tangent.
That said, the information is useful for those that don't know the differnces between PCIe 2.0 and 3.0 and what the story is with boards that claim 3.0 support. I'm well aware of the difference, thanks to talking to the OEMs directly.
Are they also found on the x58? What about the P67? How about the Z67? As I recall that is how the debate started. The Kitguru review that used the Rampage IV Extreme to provide both PCIe 2.0 vs PCIe 3.0 results. In which you wanted that board to be the only means to test PCIe 2.0 vs PCIe 3.0 results. So thanks...
I know what KitGuru did, and that's what I'm trying to point out...their results are flawed, as they aren't actually testing PCIe 3.0, regardless of what they have claimed. Other sites are claiming far different results, too
According to Intel, that makes the parts, PCIe 3.0 doesn't work. I'll listen to Intel before KitGuru. I am not going to listen to the other sites taht report on PCIe difference right now, either, so I'm not singling out KitGuru alone...
That is all. You wanting to see differences between platforms, sure, I understand, and that compare is very valid, but it won't be just PCIe that is making the difference. Myself, I'm more interesting in JUST the difference of moving PCIe 2.0 to PCIe 3.0. the comment you originally replied to was directly that..just hte pCIe 2.0 to 3.0 differences, which can be tested with a single board. WEEEEEEEEEEE!!!!
And the Rampage isn't the only board to offer that "switch"...my P9X79 Deluxe offers that as well. Review is complete, waiting for it's turn in the publishing schedule.
I don't care, it's a legit PC setup. So lets see what that does vs the other PC's.
Bad deal for the price at $549. I find it funny how people moan and cry over Nvidia on every front and AMD rewards them with $549 MSRP for a single GPU. Where is your knight in shining armor now? Financing its failed processor segment I am assuming. I should know, I have a few machines with AMD processors.
Now if they are more reasonable on the midrange they will still do fine but $549 prices them out of the reasonable category considering its probably $50 to make the card.
Separate names with a comma.