Sign in with
Sign up | Sign in
Your question

Intel Core 2 Extreme X6800 Conroe EXTREME processors atlast

Last response: in CPUs
Share
August 8, 2006 8:29:30 AM

This processor has dugged every corners and gave a good smash to AMD and currently ruling at the top.This processor has mindblowing results and none of the processor could compete it.Well it has a high price but performance wise it's the greatest

http://www.newegg.com/Product/Product.asp?Item=N82E1681...

TAKE A LOOK HERE

http://www.overclockers.com.au/article.php?id=489587

It's mindblowing

Gaming wise this processor is the topmost The games like FEAR gave the max fps
http://www.tweaktown.com/articles/914/12/page_12_benchm...

Quote:
AMD dragged away
:lol:  and well INTEL proved up to be one of the best processor maker :) 

Well would the upcoming AMD processors can give a bang to this one????
August 11, 2006 5:04:01 PM

Well being that I'm glad that Intel finally made a CPU that wasn't just good at a crunching synthetic numbers I'm kind of curious (and a bit scared at that) to see what AMD will be able to do. After all they have Hypertransport and NUMA that allow them to get buttloads of bandwidth to their CPUs and whatnot. Intel has the FSB which doesn't scale very well as we may see with quad-cores in a bit. We'll just have to see I guess.
August 11, 2006 5:16:29 PM

What exactly do you mean that "Intel has the FSB which doesn't scale very well"? It certainly doesn't seem to be holding them back greatly right now, and if you look at some of the P975 motherboards (such as the Intel D975XBX) you will see that they actually support a 1333MHz FSB, not just the current 1066MHz FSB that Conroe is using. I think they will have plenty of headroom for the forseeable future. I'll grant you that in an 8-way SMP system, AMD's HyperTransport system will have an advantage, but for most uniprocessor PCs, and even 2-way SMP, I think Intel has plenty of headroom.
Related resources
August 11, 2006 5:22:00 PM

What I mean is that right now the FSB is doing them great. I mean just look at the benchmarks that they did on anandtech between the conroe series and the AM2 CPUs, they blew them away! But here's the thing: each time you double the number of cores you divide the available bandwidth to each core in half. Right now each core has 533 MHz of bandwidth. That's plenty for now. Make it a quad-core and you get 266 MHz per core. Make it an oct-core and you get 133 MHz per core. Now given that's just a basic idea as you may not be using all the cores at the same time that's still an idea of what you're looking at. I'm not saying that Intel isn't going to come up with some way to overcome a potential bottleneck I'm just saying that it may be a problem as the amount of cores doubles that's all, and I'm not like an AMD fanboy or anything, I'm biased in either direction depending on what the benchmarks tell me.
August 11, 2006 5:26:42 PM

OMG! THIS IS BRAND NEW NEWS!! THIS TOPIC HAS NEVER BEEN COVERED ON THESE BOARDS!! THIS THREAD IS COMPLETELY NECESSARY AND INFORMATIVE AT THE SAME TIME!!!
August 11, 2006 5:28:07 PM

Wow, I looooooove your sarcasm. :roll:
August 11, 2006 5:31:44 PM

Quote:
What I mean is that right now the FSB is doing them great. I mean just look at the benchmarks that they did on anandtech between the conroe series and the AM2 CPUs, they blew them away! But here's the thing: each time you double the number of cores you divide the available bandwidth to each core in half. Right now each core has 533 MHz of bandwidth. That's plenty for now. Make it a quad-core and you get 266 MHz per core. Make it an oct-core and you get 133 MHz per core. Now given that's just a basic idea as you may not be using all the cores at the same time that's still an idea of what you're looking at. I'm not saying that Intel isn't going to come up with some way to overcome a potential bottleneck I'm just saying that it may be a problem as the amount of cores doubles that's all, and I'm not like an AMD fanboy or anything, I'm biased in either direction depending on what the benchmarks tell me.


But you're assuming that each core is going to be accessing the bus to communicate with the other, which is not necessarily the case. If you have two physically separate processors, then you get a lot of bus traffic to ensure cache consistency, as well as more general IPC. Since the cache on the core 2 processors is a shared cache, there is no FSB communication required to maintain cache consistency.

Also, I think you over-estimate the amount of bus traffic that is required in an 8-core setup. You're not going to have all 8 cores talking to the bus most of the time. It's quite possible that you would have:

Core 0: A thread handling GUI, talking with the PCI-Express graphics card and going out over the bus
Core 1: A thread handling I/O, doing networking, possibly disk I/O and keyboard I/O
Cores 2-7: Doing some calculations, preferably a small set of instructions, which will fit in the instruction cache. Practically no bus I/O if the instructions fit in the cache. The data to be operated on can be prefetched into the shared cache by thread 0 or thread 1, or pulled in directly if necessary.

Either way, someone who is thinking about writing well-threaded apps for multi-core environments can help a lot, and FSB issues will not really creep into the picture until you start talking about having multiple physical processor packages, as opposed to a single package with a shared cache.
August 11, 2006 5:32:46 PM

It's much better than yours and requires no smiley faces.
August 11, 2006 5:39:22 PM

Ahhh...

Quote:
But you're assuming that each core is going to be accessing the bus to communicate with the other, which is not necessarily the case. If you have two physically separate processors, then you get a lot of bus traffic to ensure cache consistency, as well as more general IPC. Since the cache on the core 2 processors is a shared cache, there is no FSB communication required to maintain cache consistency.

Also, I think you over-estimate the amount of bus traffic that is required in an 8-core setup. You're not going to have all 8 cores talking to the bus most of the time. It's quite possible that you would have:

Core 0: A thread handling GUI, talking with the PCI-Express graphics card and going out over the bus
Core 1: A thread handling I/O, doing networking, possibly disk I/O and keyboard I/O
Cores 2-7: Doing some calculations, preferably a small set of instructions, which will fit in the instruction cache. Practically no bus I/O if the instructions fit in the cache. The data to be operated on can be prefetched into the shared cache by thread 0 or thread 1, or pulled in directly if necessary.

Either way, someone who is thinking about writing well-threaded apps for multi-core environments can help a lot, and FSB issues will not really creep into the picture until you start talking about having multiple physical processor packages, as opposed to a single package with a shared cache.


Good point... thanks for educating me (no sarcasm in there, seriously)

Now what happens when you start doing heavy multi-tasking? I only ask this becuase I do that A LOT!
August 11, 2006 6:11:24 PM

It depends on what you consider to be heavy multi-tasking, but suffice to say that it will likely degrade performance. Basically, the way that the processors are designed are as follows (for the core 2):

Each core has its own L1 cache, which is 64KB per core. 32KB of this is for data, and 32KB are to cache instructions. The data cache between the cores must be consistent - i.e. if both cores are working on the same memory segment, there is communication between the cores to make sure that each has the correct data when necessary. I haven't read enough of Intel's whitepapers to know if they are doing this internally, or if they are being lazy and going out on the FSB to accomplish this, but I'm getting off track.

The important part is that because each core has its own L1 cache with its own Icache (instruction cache - 32KB), each core can cache its instructions for its running thread without them getting bumped out by the thread running on the other core. (Of course, if you have more threads than cores, which you always do except in embedded processing cases, threads will always be swapped in and out, but this happens relatively infrequently with respect to the speed at which instructions are executed).

The most critical factor is whether or not the thread fits within the L1-cache - especially true for the instructions, but also important for the data. (I'm not sure if the L2-cache is data only or if instructions can also be stored in the L2 cache - again, I haven't read Intel's whitepapers on Core 2). If you have two threads that both fit within their respective core's cache, for the most part (with the occasional miss having to go to the L2 cache) you will still be operating very fast. If you have more high-priority threads than cores, you will have threads getting kicked out, and this will obviously slow you down.

I don't know how much that really answers your question, but I tried.
August 11, 2006 6:16:01 PM

Quote:
Gaming wise this processor is the topmost The games like FEAR gave the max fps


103 for the fx-62 and 107 for the E6800 on FEAR.... not much of a difference. It only had a diffence of about 300 points in 3Dmark06. However, it did have very impressive results in Q4, Far cry, and COD2. It all depends on how much you want to spend.
August 11, 2006 6:22:50 PM

Quote:
Gaming wise this processor is the topmost The games like FEAR gave the max fps


103 for the fx-62 and 107 for the E6800 on FEAR.... not much of a difference. nobody would know the difference.
Wherever you're getting those numbers from, they're most likely GPU limited.
August 11, 2006 6:23:08 PM

No you did a good job explaining it. Heavy multi-tasking for me: email, trillian, watching a DVD while doing a defrag and waiting for Maya to finish rendering something. I'm glad I'm going with a woodcrest system though.. more overall CPU power.
August 11, 2006 6:26:00 PM

If they're GPU limited, then people really aren't going to notice a difference between the two processors :-)

Seriously though, my monitor's native resolution is 1920x1200. Practically any game I play (except for perhaps Warcraft 2: Tides of Darnkess) is going to be GPU limited at that resolution.
August 11, 2006 6:29:33 PM

Quote:
Wherever you're getting those numbers from, they're most likely GPU limited.


http://www.tweaktown.com/articles/914/12/page_12_benchm...

According to this site, it the benchmark was run at low resolutions to shift the load fro GPU to CPU.
Yeah, I don't know how much stock I'd put in those benchmarks. It has the Pentium EE 955 beating all of them in Quake 4.
August 11, 2006 6:31:04 PM

Exactly... and I really don't know why you would want to run at a low resolution. Perhaps if you're using an older LCD monitor that only supports 1024x768 (or 1280x1024 even...), but if you're running 1600*1200 or 1920*1200, you're definitely going to be GPU limited on any of the recent games.
August 11, 2006 6:50:56 PM

Quote:
Exactly... and I really don't know why you would want to run at a low resolution. Perhaps if you're using an older LCD monitor that only supports 1024x768 (or 1280x1024 even...), but if you're running 1600*1200 or 1920*1200, you're definitely going to be GPU limited on any of the recent games.


which also probably makes the difference in Frames per second minimal?
August 11, 2006 6:56:36 PM

Quote:
Exactly... and I really don't know why you would want to run at a low resolution. Perhaps if you're using an older LCD monitor that only supports 1024x768 (or 1280x1024 even...), but if you're running 1600*1200 or 1920*1200, you're definitely going to be GPU limited on any of the recent games.


which also probably makes the difference in Frames per second minimal?

Difference in FPS between the two processors? Yes, quite minimal, because it's not the processor that's the bottleneck, it's the GPU.
August 11, 2006 7:01:13 PM

Impressive.... I can't wait to see how AMD responds to this, and if their new 4x4 or k8L will be enough to beat the conroe
August 19, 2006 3:11:46 PM

Well I dont think at all that the x6800 can be beaten by any of the processors.It's like the release of the ATI x1900xt which created a huge chaos in nvidia and then with no future plans nvidia released the 7950GTX and 7950GX2 with 1GB DDR3


[code:1:fe906421fe]
AMD must seek the pathway to god[/code:1:fe906421fe] :D 
August 19, 2006 3:25:55 PM

What's a processor?
August 19, 2006 3:39:03 PM

You damn sure an AMD fanboy becoming angry
August 19, 2006 3:58:35 PM

If only you hadn't posted old news. :?
!