Compatibility worries

Well, this thread has become all about semantics anyway, so thanks for the additional input, I hope it doesn't create a "bottleneck"...! :rolleyes: :haha:

I took it as constipated.....oh well tomato tomato,
they are correct however, most people have 'bottleneck' 180 degrees from what it is.

anyway Chef/Cap so to put this in context, using Metro 2033 as an example of a game that brings most VGA's to it knees, is the game "unoptimized" or is it creating a 'bottlneck'from the number of textures/triangles and or physics? or for that matter both?

The graphical features are the real killer -tessellation and depth of field impact performance severely. Running the game on AMD hardware is going to have a further penalty, since the game is optimised for GPU PhysX, which I think is offloaded (poorly.. x87 compile?) to the CPU with AMD cards.
I have the game installed on my secondary rig (HD5850's) at the moment and I wouldn't say the performance/IQ is anything to write home about. My primary rig (which is on semi-permanent loan to my nephew) also has the game loaded and runs like melted butter using the GTX 280's (with higher IQ)...but then, the game was optimised to use a dedicated thread for PhysX, so maybe with AMD GPU's, two (or more) threads are seconded to do the physics work.
I think the game could have been coded better-although that is coloured from the comments of others, but that's guesswork as I would need to see another game using the 4A engine to judge.

There's a bunch of different interpretations. I would offer this, software that only utilizes one thread can't "bottleneck" a CPU. The CPU outputs that which it can, generating the throughput it is capable of. That's not a bottleneck, that's a part in need of an upgrade. However, perhaps bottleneck is more appropriate in a software environment that has multi threading that the processor can't deal with. I simply don't equate a part incapable of sufficient throughput as a bottleneck, only a part that limits the output of the previous link in the chain. For example, my 50 FPS VGA installed after a 200 FPS CPU analogy. Perhaps my definition is too narrow, perhaps I like to argue, perhaps so does everybody else.

With that said, "bandwidth"is also a term slightly misapplied in certain areas. Bandwidth is only appropriate if information is being processed in parallel, in some cases, "band speed" might be a better adjective. "And back to you in the studio Red...."

"Bandwidth" is a term that was coined in conjunction with TV/radio transmission/ receiving technology. In the case of AM transmission, Bandwidth would refer to the Peak to peak differential of signal modulation, whereas, with FM transmission, "bandwidth" refers to the frequency differential of the signal, and it's relationship to interference with adjacent channels. Both statements could be true here also, if the IP were processing information from many clients simultaneously, in parallel, then bandwidth would apply, if information is only being serially dispersed the "stream speed" would be more appropriate.

okay then...Thanks Cap....hot enough out there for ya?
well I get exactly where you are coming from, this is why you are getting the feedback, I think, because this is what is largely being offered as the definition of bottleneck. It is that it is simply the slowest and most incapable component in the system

http://www.youtube.com/watch?v=x6hkGiYW21s

However if you read reviews/benchmarks at Tom's for example, the 'bottleneck is the component that cannot, or is not, being used to its full potential. and its being caused by the program that is being run.

I am with you on this. to me this is a 'weak link' not a bottleneck. However Ihave to stop and count fingers to make sure which I am talking about at times, because they are different.

Here What I do to Stop Counting My Fingers......

I feel you with that. I put my free hand in my pocket while I'm online, although just not at Techspot. In a serendipitous turn of events, I'm left handed, but I still mouse with my right. :rolleyes:

....and I knew you would take a shot at that when I typed it :haha:

...that have anything to do with your new monitor?

I'm going into "Disc Management" now. I'm going to try and rename the "G:/" drive, the "OY:/" drive.

Well no, not exactly. That is on a machine not connected to the internet. When I use that, I give it my undivided attention, which means, "both hands"!

of course, what a fool Ive been!
I just moved, maybe you can see the smoke signals from my house now.:rolleyes:

There's an outside..? Wow, what a fool I've been...!

I'm actually afraid to go out and look, the natives have been really, really restless today.

"Not so pale puss have'm many guns"

Well there was. It was allegedly taken sometime last night and is reportedly being held hostage in an apartment outside New Jersey.

No, I rather think my neighbors sold it to buy crack. Unless they're really going to go through with a "Sopranos" reunion movie. In which case Vinnie has it, or was he rubbed out?

This is getting punchy, even for us. :haha:

well if it helps brighten your lousy sounding day, I broke two ribs in the move, and it hurts like hell when I laugh. You have inflicted a sufficient amount of pain in this thread.

Now Red, didn't anybody ever tell you that moving is not a full contact sport? A little voice in your head should have been screaming, "get out of the way of the falling piano", long before it ever hit you....!

On that note, I'll wish you a goodnight. BTW, if the pain goes away before the meds do, save a buddy enough to get him high.

It was an anvil, and I didn't even know the roadrunner who dropped it.

Wow you guys are great! Thanks for all the info especially the CPU comparison write up! AWESOME! Ok just so you guys know how this came about, I strongly believe that it is possible to produce a 1 core CPU that is capable of putting out what a now day i7 is scoring! (but that's for another posting and another time!) I also believe that this issue of bottle-necking has alot more to do with software and drivers! What I did when my friends (making fun) told me that my computer was a bottleneck itself was show them a little test! (Please explain this and I will be a firm believer in hardware bottle-necking.) I run a dual display, so on one screen I bring up the EVGA GPU utility and Windows Vista(32-bit Edition) Task manager performance screen. I crank up Crysis demo version downloaded from Nvidia's site. During the game I get 100% GPU utilization and a 73% CPU utilization that max out at about 81 or 82% and sometimes drop as low as 60% during light(graphical) moments in the game. The logical core usage fluctuates to sometimes mimic the physical core but it never exceeds the 81 or 82% usage. If hardware bottle-necking is present shouldn't it be the other way around where the CPU is at 100% and the GPU would be breezing through the calculations the CPU is feeding it with very minimum GPU utilization? Now say I added a dual-core, what more can I expect from my GPU if its usage was already maxed by a single core? Captaincranky, I believe you said something similar in one of your postings under this subject about the CPU still being the main one doing all the calculations and I agree with you 150%! Anyways thanks in advance for all of you guys input! I have emailed Intel a similar question but they never seem to reply! Thanks Again! (Also try the little test out yourself!)

Assuming that you are talking about P4 here, because that was the last mainstream desktop processor to have one physical core from Intel, before Pentium D's came out.

Now that out of the way, P4 (Prescott and later) had very deep pipeline, i.e. 31 stages (if I remember correctly), having such a long pipeline makes it easier to jack up the speeds, but the downside is, accurate branch prediction becomes extremely critical, because whenever a misprediction occurs, it would result in processor halting and clearing out the whole pipeline before starting over again. However, one important milestone of P4 (Northwood onward) was introduction of HyperThreading (which is also called Simultaneous Multi-Threading i.e. SMT).

Nehalem has a very wide pipeline. It has six execution units capable of executing three memory operations and three calculation operations. If the execution engine can’t find sufficient parallelism of instructions to take advantage of them all, “bubbles”—lost cycles—occur in the pipeline.

To address this issue, SMT looks for instruction level parallelism in two threads instead of just one, with the objective of leaving as few units unused as possible. This approach can be extremely effective when the two threads are executing tasks that are highly separate. On the other hand, two threads involving intensive calculation, for example, will only increase the pressure on the same calculating units, putting them in competition with each other for access to the cache. It goes without saying that SMT is of no interest in this type of situation, and can even negatively impact performance.

Generally, the impact of SMT on performance is positive most of the time and the cost in terms of resources is reasonably limited, which explains why Intel brought it back with i7. However, do note that programmers will need to pay attention because with Nehalem (and its later incarnations), all threads are not created as equals. To help solve this conundrum, Intel provided a way of precisely determining the accurate topology of the processor (the number of physical and logical processors), and programs can then use the OS affinity mechanism to assign each thread to a processor.

Since SMT puts a heavier load on the out-of-order execution engine, Intel has made significant improvements in this area as well on i7 and increased the size of certain internal buffers to avoid turning them into bottlenecks. So the reorder buffer, which keeps track of all the instructions being executed in order to reorder them, has increased from 96 entries on the Core 2 to 128 entries on Nehalem. In practice, since this buffer is partitioned statically to keep any one thread from monopolizing all the resources, its size is reduced to 64 entries for each thread with SMT.

Obviously, in cases where a single thread is executed, it has access to all the entries, which should mean that there won’t be any specific situations where Nehalem turns out to have worse performance than its predecessor.

Therefore, you can't really imply that generation x processor's core will perform on the same level as the generation y's processor core.

Sources: For i7, Toms, I forgot where I read architectural details about Prescott's since it was light years ago