Queries about the dual revolution

[MetalheadGautham]

I have lots of questions about the new dual giri in computers. they are:

1. Does a XGhz dual core processor with Ymb L2 cache have any advantage over a 2XGhz single core processor with the same Ymb L2 cache? If yes, WHY? Assume both are of the same quality.

2. Does a X+X ram configuration have any advantage over a single 2X ram configuration? if yes, WHY?

3. Does a similar thing in graphics card also show any advantage? why?

4. Is an X mb ram running at 2Ymhz better than a 2X mb ram running at Y mhz? why?

5. how advantageous is it to have an L3 cache?

6. is an onboard graphics card with a set of features and a certain ram alloted to it worser than a dedicated graphics card with the same set of features and with the same amount/type of ram? why?

7. Why do costs of processors go up with the amount of L2 cache? what difference does it exactly make? And what is the approximate cost per mb of L2 cache?

Note: assume some values for X and Y. if x=2, 2x=4. you get my point...

[gxsaurav]

Originally Posted by MetalheadGautham

1. Does a XGhz dual core processor with Ymb L2 cache have any advantage over a 2XGhz single core processor with the same Ymb L2 cache? If yes, WHY? Assume both are of the same quality.

No, 2 physical processor are always better then a dual core CPU.

2. Does a X+X ram configuration have any advantage over a single 2X ram configuration? if yes, WHY?

Yup, if running in Dual Channel mode as it doubles the total bandwidth of memory.

3. Does a similar thing in graphics card also show any advantage? why?

Yup, u mean SLI. In this case usually drivers go for alternate frame rendering or Stacked rendering. Means card 1 is rendering the frame u see right now, while card 2 is already rendering the next frame...due to this the overall frame rates increse resulting in a smoother gameplay.

SLI GPUs also for for partial or stacked rendering, in this case Card 1 renders half the frame while card 2 renders the rest of the frame at the same time.
4. Is an X mb ram running at 2Ymhz better than a 2X mb ram running at Y mhz? why?

5. how advantageous is it to have an L3 cache?

The CPU will not need to go to system RAM for it's tasks. There is more memory available for the CPU in form of L3 cache resulting in rapid execution & storage/flush of data

6. is an onboard graphics card with a set of features and a certain ram alloted to it worser than a dedicated graphics card with the same set of features and with the same amount/type of ram? why?

Yup, it's worse. 1st it's sharing RAM with system which is always slower then dedicated memory of a graphics card which usually runs at much higher bandwidth. Like the Memory bandwidth of my gfx card is about 14 GB while that of my my System RAM is 4 GB/sec

7. Why do costs of processors go up with the amount of L2 cache? what difference does it exactly make? And what is the approximate cost per mb of L2 cache?

Cost, don't know. Adding more cache means a complex manufacturing process & more transisters resulting in higher price of CPU.
Note: assume some values for X and Y. if x=2, 2x=4. you get my point...[/quote]

[The_Devil_Himself]

Originally Posted by MetalheadGautham

5. how advantageous is it to have an L3 cache?

L3 cache is cheaper and slower than L2 cache iirc and it is used by AMD to lower the costs of their processors.

[Choto Cheeta]

1. Does a XGhz dual core processor with Ymb L2 cache have any advantage over a 2XGhz single core processor with the same Ymb L2 cache? If yes, WHY? Assume both are of the same quality.

subjective.. how can both be same quality ?? there are 2 processor in Dual Core, and single on Single, then how can they be same quality ??

2. Does a X+X ram configuration have any advantage over a single 2X ram configuration? if yes, WHY?

what are u refering is to Dual channel RAM config...

with X+X config that 2X, it will double the amount of available memory bandwidth....

3. Does a similar thing in graphics card also show any advantage? why?

yes, but again subjective as not many software / applciation ro rather games are in main stream which can take effect of this...

But for ur info, once again, under SLI / Crossfire, it will double the amount of available memory bandwidth.... and also doubling the Graphics processing power...

once again there are very few main stream apps which can take advantage... off course advance medical simulation or such do take heavy advantage of Pulti GPU (Quad SLI)

4. Is an X mb ram running at 2Ymhz better than a 2X mb ram running at Y mhz? why?

MHz would be how fast the application would be processed and MB would mean how much of the apps would be processed...

so once again the question is subjective...

higher clock speed means higher bandwidth means faster processing, but if u have a apps which needs more ammount of memory to use for it self then though u have higher bandwidth but it would slow u down...

For example... if u are running vertual Applicatioons then more MB would matter, but if u are editing video / encoding then after certain amount of MB, their bandwidth would matter

5. how advantageous is it to have an L3 cache?

L3 ?? Processor u mean to say ?? if yes, then 1st read what processo Cache memory is for...

http://en.wikipedia.org/wiki/CPU_cache

on short its the fasted RAM on ur system... though its little... but its used for handeling the most used instuction set without taking it to ur RAM for faster processing...

So yes extra ammount of Cache will help a lot...

6. is an onboard graphics card with a set of features and a certain ram alloted to it worser than a dedicated graphics card with the same set of features and with the same amount/type of ram? why?

If the bold part is to be consider then once again it becomes subjective,,,

for example, your consider MX 4000 or such card of nvidia and compare a Intel G965 / G33 or nvidia 6150 or AMD 690G

you would see its running better than dedicated card...

7. Why do costs of processors go up with the amount of L2 cache? what difference does it exactly make? And what is the approximate cost per mb of L2 cache?

Refer to question 5




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

I am not the best person for computer Architecture... home some one from this forum would give you better answers

[a_k_s_h_a_y]

i don't know where L2 Cache is really put into the new age computer architecture (on old comps its on mobo)
am soo confused if L2 is in procy or mobo

L1 Cache Matters a hell lot and is directly built into Processor

Cache is Made of Registers and each Register consist of many transistors and diodes and they are Static Memory Device .. Data in cache is available ready for use at any time

Where as in Ram each unit is made of 1 transistor and 1 Capacitor !
And RAM Is Dynamic Memory . Capacitor has to be continuously recharged
and data access takes place only at given times .. so data available only when capacitor is full and not while charging

hence RAM Is Slow ! Cache (L1,L2,L3) is way too faster then RAM

Since Register Design is Complex and needs more resources More Cache means more money

[sivarap]

Nice explanation choto....

(knew em already...but the presented way is good)

[Choto Cheeta]

@sivarap

thnx....

[sivarap]

Originally Posted by a_k_s_h_a_y

i don't know where L2 Cache is really put into the new age computer architecture (on old comps its on mobo)
am soo confused if L2 is in procy or mobo

L1 Cache Matters a hell lot and is directly built into Processor

Cache is Made of Registers and each Register consist of many transistors and diodes and they are Static Memory Device .. Data in cache is available ready for use at any time

Where as in Ram each unit is made of 1 transistor and 1 Capacitor !
And RAM Is Dynamic Memory . Capacitor has to be continuously recharged
and data access takes place only at given times .. so data available only when capacitor is full and not while charging

hence RAM Is Slow ! Cache (L1,L2,L3) is way too faster then RAM

Since Register Design is Complex and needs more resources More Cache means more money

oh... never knew this....thanks.

also the faster access is also due to the transmission distance/medium....(limited to FSB frequency when using the ram) hope i am right.

[New]

Nice explanation guys..

[MetalheadGautham]

I think you misunderstood me a bit.

in dual core, I ment something like

1.6Ghz each core, 1mb L2 Cache say intel dual core processor vs 3.2Ghz 1 mb L2 cache P4 with HT or something.

================================================== ==========

in ram I ment something like two 1 GB rams vs one 2 GB ram of the same say 1066

Mhz frequrncy.

================================================== ==============

in graphics cards I did not ask about SLI mode.

I clearly stated same system resources. ex: XYG4398 onboard with 1.25 gb ram on pc(0.25 alloted for graphics card) with ram frequency say 1066 MHz with pixel shader 2.0, antialiasing, etc vs ABX420 dedicated with 1 gb ram on PC at 1066 MHz, 256 mb ram in graphics card running at 1066 MHz, with pixel shader 2.0, antialiasing, etc. I said same resources/features, so I thought you guys understood properly.

================================================== ==============


as for cache, I read that its possible to have 256 MB L3 in a PC. is it true? and I still did not get the answer to how important L2 cache is or how costly it is. I mean, if it is much faster and only slightly costlier than dynamic ram, won't all buy Static RAMs?

================================================== ==============

anyway, thanks for the replies saurav and cheeta...

[xsreality]

Originally Posted by MetalheadGautham

I think you misunderstood me a bit.

in dual core, I ment something like

1.6Ghz each core, 1mb L2 Cache say intel dual core processor vs 3.2Ghz 1 mb L2 cache P4 with HT or something.

There is no comparison b/w C2D proccys and the old Pentium D/P4 proccys... C2D are way faster! And don't confuse with those GHz speeds.. 3.2GHz of Pentium D is nothing in front of 1.6GHz of C2D... Even two Pentium D processors working in parallel can't beat the entry level C2D processor.i hope u get it..

in ram I ment something like two 1 GB rams vs one 2 GB ram of the same say 1066

Mhz frequrncy.

Others meant the same thing... former is faster.

in graphics cards I did not ask about SLI mode.

I clearly stated same system resources. ex: XYG4398 onboard with 1.25 gb ram on pc(0.25 alloted for graphics card) with ram frequency say 1066 MHz with pixel shader 2.0, antialiasing, etc vs ABX420 dedicated with 1 gb ram on PC at 1066 MHz, 256 mb ram in graphics card running at 1066 MHz, with pixel shader 2.0, antialiasing, etc. I said same resources/features, so I thought you guys understood properly.

Again, keeping the system resources same, 2 grfx card working together means SLI/Crossfire...

as for cache, I read that its possible to have 256 MB L3 in a PC. is it true? and I still did not get the answer to how important L2 cache is or how costly it is. I mean, if it is much faster and only slightly costlier than dynamic ram, won't all buy Static RAMs?

not much idea abt that... 256MB L3 in a PC though is something of a supercomputer kinda thing.. mayb others might shed more light on that... L2 cache is definetly much faster and its not just slightly costlier... but cost of a cache makes no sense as it is not something to be bought in the market.. u get it in fixed sizes with various proccys..

[MetalheadGautham]

where did I ask about SLI? I asked about an Onboard Vs a Dedicated with the configuration I gave... where does it even slightly indicate SLI?

[assasin]

Originally Posted by MetalheadGautham

I think you misunderstood me a bit.

in dual core, I ment something like

1.6Ghz each core, 1mb L2 Cache say intel dual core processor vs 3.2Ghz 1 mb L2 cache P4 with HT or something.

u cant compare C2Ds with Penrium Ds cuz dont hav the same architecture.

as for ur q,if both had same architecture,the dual core with 1.6Ghz will perform better for apps which supports multi core processors cuz the dual core proccy is able to execute 2 threads simultaneously.if the apps doesnt support multicore proccy then the 3.2Ghz wil perform better.
same is the case with C2D E6850 and C2Q Q6600.E6850 performs better or is equivalent in apps which dont support quad core proccies.

also apps which r more dependent on clk speed will perform better on proccies with higher clk speeds.

remember all this applies only if u assume thatboth the single core and the dual core processors hav the same architecture.

[MetalheadGautham]

Originally Posted by assasin

u cant compare C2Ds with Penrium Ds cuz dont hav the same architecture.

as for ur q,if both had same architecture,the dual core with 1.6Ghz will perform better for apps which supports multi core processors cuz the dual core proccy is able to execute 2 threads simultaneously.if the apps doesnt support multicore proccy then the 3.2Ghz wil perform better.
same is the case with C2D E6850 and C2Q Q6600.E6850 performs better or is equivalent in apps which dont support quad core proccies.

also apps which r more dependent on clk speed will perform better on proccies with higher clk speeds.

remember all this applies only if u assume thatboth the single core and the dual core processors hav the same architecture.

I did not compare C2D and dual core. I compared Pentium 4 with HT technology against C2D or some similar processor with both having same net clock speed and cache.

and that means mono core VS dual core

and I never even said Pentium D.(when I say Dual Core, I mean just that. Two Cores. Not any specific model of processor)

btw, I heard Pentium 4 with HT, Intel's last single core Processor, supports multithreading. so the comparison IS fare here assuming both have the same clock speeds.

won't that mean that in some single thread apps, which require lots of processing power(more than half the total available power) they will not run on 1.6 GHz two cored procy but will run on 3.2 GHz multi thread compatible one cored procy?