Would a P4 3.2GHz bottleneck with a 7900GT?

Status
Not open for further replies.
There's 2 different 3800's. There's a single core clocked at 2.4 with 512mb of cache. He's saying don't buy THAT one.

The dual core runs at 2ghz with 2 sets of 512mb of L2 cache.

Ryan
 
A processor does not bottleneck a video card as there is not direct communication between the two...all P4s would rely on either an 800 or 1066MHz bus to communicate with the video card and other components, and all amd64s would rely on the 2000MHz HTT bus for point to point communication. The bus speed doesn't change regardless of whether the internal operating frequency of the core does so I have no idea where you guys get off saying there is any sort of bottleneck present

Now obviously people are going to come in here and tell me an FX-57 will get x number of frames more than a 3000+, but that's because you're changing the instruction cycle period on the CPU meaning it's getting ITS workload done a little bit faster, but that isn't going to have any bearing on how the GPU performs. Adding a faster CPU and calling that fixing a bottleneck would be no different than having a 6800GT with an FX-57 and adding a 7800GT and calling that fixing a bottleneck...all you're doing is adding a faster INDEPENDANT component to your machine
 
Thank you gaara. It's so funny how people are always shouting "CPU Bottleneck" with relation to GPU > Rendering.Display

If people are worried about bottlenecks, they should first look at their software, which is the #1 bottleneck in systems today.

The Second, hardware bottleneck is currrent Harddrives. Compared to other components like CPU & RAM, HDD's are slower then molasses in January. (Even SCSI HDD's)

"Will my WD Raptor 10K 150G HDD bottleneck my OCZ PC6400?"

The answer is absolutely. Even PC2100 will be bottlenecked.
 
Spartan666 said:
umm...no, ur wrong. the X2 3800+ is 2.0Ghz has 2*512kb L2 cache, and what are your reasons for not recommending it??

john3 said:
yeah seriously maybe u should do a lil research......the 2.4GHz is the 4400+, right? and i think the 3800+ is a good CPU, u can OC it 2 about 2.5, right? if you no wat ur doing.

Yea, you're telling ME to do my research? You're the idiot for calling it a 3800+. There is a pretty big difference between the 3800+ and X2 3800+ you dipshit. Learn to type and go do your research.

gaara said:
A processor does not bottleneck a video card as there is not direct communication between the two...all P4s would rely on either an 800 or 1066MHz bus to communicate with the video card and other components, and all amd64s would rely on the 2000MHz HTT bus for point to point communication. The bus speed doesn't change regardless of whether the internal operating frequency of the core does so I have no idea where you guys get off saying there is any sort of bottleneck present

Now obviously people are going to come in here and tell me an FX-57 will get x number of frames more than a 3000+, but that's because you're changing the instruction cycle period on the CPU meaning it's getting ITS workload done a little bit faster, but that isn't going to have any bearing on how the GPU performs. Adding a faster CPU and calling that fixing a bottleneck would be no different than having a 6800GT with an FX-57 and adding a 7800GT and calling that fixing a bottleneck...all you're doing is adding a faster INDEPENDANT component to your machine

Thanks for clearing that up. Nubius also tried to explain it, but I get it now. I won't say that a GPU bottlenecks the CPU, but how do you explain crappy processors holding back video cards?

HAVOC2k5 said:
Thank you gaara. It's so funny how people are always shouting "CPU Bottleneck" with relation to GPU > Rendering.Display

If people are worried about bottlenecks, they should first look at their software, which is the #1 bottleneck in systems today.

The Second, hardware bottleneck is currrent Harddrives. Compared to other components like CPU & RAM, HDD's are slower then molasses in January. (Even SCSI HDD's)

"Will my WD Raptor 10K 150G HDD bottleneck my OCZ PC6400?"

The answer is absolutely. Even PC2100 will be bottlenecked.

That, however, made no sense. You've got to take into account real-life performance. How do you explain the increase in performance going from PC2100 memory to PC6400 memory? Nope, not a bottleneck there. And not the software either. You've got do back up your stuff like Gaara did, man.
 
Green Radience said:
Yea, you're telling ME to do my research? You're the idiot for calling it a 3800+. There is a pretty big difference between the 3800+ and X2 3800+ you dipshit. Learn to type and go do your research.

You're already hanging yourself dude... I'd tone it down fast. ;) I hope you know exatly what I'm saying.

Ryan

Rearrange the underlined letters to spell the secret word...
 
well i didnt really say i was getting the 3800 single core, so w/e.

lol.

Flanker?

oh, my bad.
 
I won't say that a GPU bottlenecks the CPU, but how do you explain crappy processors holding back video cards?
They aren't and that was my point, a GPU is an independant component hence why it has it's own core and dedicated memory, therefore no other component has any bearing on how fast it operates.

I'll try expressing it as a formula. The number of instructions a GPU executes per cycle can be represented by x and the number of instructions executed per cycle by the CPU can be represented by y.

Framerate = x + y

It's that simple, and you can see that they're seperate. A CPU and GPU are both responsible for their own processes when you're playing a game, and the only time you they communicate with another is to assemble the data together which of course is reliant on your bus speed which as I mentioned doesn't change regardless of internal frequencies. You can see that a GPU will always continue to render x number of instruction sets and the CPU will always continue to process y number of instruction sets, and you add them together to get your framerate

So a crappy CPU doesn't hold back the GPU, it only holds itself back in that sense considering it's independantly responsible for its own set of tasks

That, however, made no sense. You've got to take into account real-life performance. How do you explain the increase in performance going from PC2100 memory to PC6400 memory? Nope, not a bottleneck there. And not the software either. You've got do back up your stuff like Gaara did, man.
I believe he's talking about the page file or just pulling data from the hard drive into your memory in general...in which case he's right considering hard drive bandwidth is probably limited to about 70mb/sec on most average 7200RPM drives, comparing to 6400mb/sec memory bandwidth
 
They aren't and that was my point, a GPU is an independant component hence why it has it's own core and dedicated memory, therefore no other component has any bearing on how fast it operates.

I'll try expressing it as a formula. The number of instructions a GPU executes per cycle can be represented by x and the number of instructions executed per cycle by the CPU can be represented by y.

Framerate = x + y

It's that simple, and you can see that they're seperate. A CPU and GPU are both responsible for their own processes when you're playing a game, and the only time you they communicate with another is to assemble the data together which of course is reliant on your bus speed which as I mentioned doesn't change regardless of internal frequencies. You can see that a GPU will always continue to render x number of instruction sets and the CPU will always continue to process y number of instruction sets, and you add them together to get your framerate

So a crappy CPU doesn't hold back the GPU, it only holds itself back in that sense considering it's independantly responsible for its own set of tasks

---------------------------------------------------------------------------------------

Thank you, that makes perfect sense and seem very logical, now I understand that I was wrong to ask the topic question in the first place
 
gaara said:
I believe he's talking about the page file or just pulling data from the hard drive into your memory in general...in which case he's right considering hard drive bandwidth is probably limited to about 70mb/sec on most average 7200RPM drives, comparing to 6400mb/sec memory bandwidth

wow. I thought that was common knowledge and didn't have to explain it like gaara did. I guess for now on i'll have to list every last detail with bullets and put things into a neat little outline.

gaara may feel like teaching you guys, But I certainly don't.
 
Status
Not open for further replies.
Back
Top Bottom