Hi Tim & Richard
Great review!(as usually) and +1 for including results of overclocked CPU.
Seems like AMD should focus on core architecture to catch up with intel(or leap ahead:)).
I've got one question for you - I've bought Athlon X2 5000+ black ed., Asus M3-A32 MVP dlx and Corsair XMS2 at 800MHz CL4.
I managed to overclock it from 2.6GHz to 3.2GHz. I only had to rise CPU voltage to 1.325V to get it running completely stable, however I had to loosen memory timings from 4-4-4-12-2T to 5-5-5-15-2T. I would like to get mem's running at CL4 setting. Can you give me any advice on this? There are some other voltages to increase, but I'm not sure about them.
I'll be very thankful for any help.
We were told and were expecting the CPU to be £90-100: basically an E6550 contender. However we waited until this morning and it's far in excess of the price we were told :(
Have you increased the memory voltage/northbridge voltage (if there's a specific option for the CPU-NB)?
Have you increased the memory voltage/northbridge voltage (if there's a specific option for the CPU-NB)?
That was quick!
It's the very same mobo you used(M3A32) - I think I noticed "northbridge voltage" there.
I'll try it(and perhaps report how much it helped:))
Thanks!!!
I don't know what I'm most disappointed about: the lack of higher clock speeds, the stupid model number system AMD's using which stops them raising the clock speeds, AMDs slides saying quad was upto 20% faster than tri when we can clearly see it's >20% faster at times or the fact that Bit-tech didn't try gaming and encoding at the same time. Now so that you don't think I'm completely unappreciative of the work you do: Thank you, a nice thorough review as always.
Originally Posted by Kúsař That was quick!
It's the very same mobo you used(M3A32) - I think I noticed "northbridge voltage" there.
I'll try it(and perhaps report how much it helped:))
Thanks!!!
"Northbridge" is different to "CPU-NB". I know Phenoms have the CPU-NB option which is specific to the memory controller on the M3A32
I saw the numbers and X3 and thought, what kind of crazy graphics card is this for a second. I hope AMD can get a good high end chip into the market soon, competition is so much better for the consumer, like you don't know that though.
I have a question about AMD and ATI products. ON paper their Spec out perform the Nvidia GPUs...
400 or more stream proccessors compared to the 128 or 256. Its really wierd. Higher clock speed etc. Whats going on. They even use ddr4 and still cant beat the GTX versons of high end cards. there X2 came yes it was powerful on paper and couldnt out perform the gx2
Originally Posted by xtremeownage I have a question about AMD and ATI products. ON paper their Spec out perform the Nvidia GPUs...
400 or more stream proccessors compared to the 128 or 256. Its really wierd. Higher clock speed etc. Whats going on. They even use gddr4 and still cant beat the GTX versons of high end cards. there X2 came yes it was powerful on paper and couldnt out perform the gx2
DAMMIT (AMD/ATI) and Nvidia have different ways of counting stream processors:
AMD counts a cluster of 4 small ones (which can't do special functions) and 1 fat one (which can do special functions) as a single stream processor.
Nvidia counts a cluster of one normal processor and one special function processor as a single stream processor.
- note that neither processor is comparable to its competitor's version.
So the total number of individual processors in:
AMD's Radeon HD 3870 - 320 SP running at 775 MHz - Total GigaFLOPS = 496. Core clock speed = 775 MHz. Mem speed = 2250 MT/s GDDR4.
Nvidia's GeForce 9800 GTX - 256 SP running at 1125 MHz - Total GigaFLOPS = 416. Core clock speed = 650 MHz. Mem speed = 970 MT/s GDDR3.
- Reference specifications. The core clock speed is what the rest of the chip (i.e texture units, command processor, output units etc.) runs at. Since both cards I've mentioned have identical ammounts of RAM and an equally wide bus to RAM, their RAM bandwidth is directly related to the speed of their RAM. GDDR4 has higher latency than GDDR3 so although it runs faster it has to do this to 'catch up', it offsets this by being able to run at higher speeds.
But the real reason for the Radeons falling behind the GeForce is the difference in texturing capacity: 12.4 GigaTexels verses 41.6 GigaTexels. I'm sure that the fact Nvidia works with almost every game developer has nothing to do with. .
Comments 1 to 19 of 19
Great review!(as usually) and +1 for including results of overclocked CPU.
Seems like AMD should focus on core architecture to catch up with intel(or leap ahead:)).
I've got one question for you - I've bought Athlon X2 5000+ black ed., Asus M3-A32 MVP dlx and Corsair XMS2 at 800MHz CL4.
I managed to overclock it from 2.6GHz to 3.2GHz. I only had to rise CPU voltage to 1.325V to get it running completely stable, however I had to loosen memory timings from 4-4-4-12-2T to 5-5-5-15-2T. I would like to get mem's running at CL4 setting. Can you give me any advice on this? There are some other voltages to increase, but I'm not sure about them.
I'll be very thankful for any help.
Cheers
-K
We were told and were expecting the CPU to be £90-100: basically an E6550 contender. However we waited until this morning and it's far in excess of the price we were told :(
Have you increased the memory voltage/northbridge voltage (if there's a specific option for the CPU-NB)?
That was quick!
It's the very same mobo you used(M3A32) - I think I noticed "northbridge voltage" there.
I'll try it(and perhaps report how much it helped:))
Thanks!!!
And is it just me, or is the colour key for the "PCMark Vantage x64 Composite Scores" still off like it was in the last review?
"Northbridge" is different to "CPU-NB". I know Phenoms have the CPU-NB option which is specific to the memory controller on the M3A32
Amon - "off"?
Yep, the problem is price - nothing else, really. :)
pitted against a BE is it a viable option for the price??
400 or more stream proccessors compared to the 128 or 256. Its really wierd. Higher clock speed etc. Whats going on. They even use ddr4 and still cant beat the GTX versons of high end cards. there X2 came yes it was powerful on paper and couldnt out perform the gx2
AMD counts a cluster of 4 small ones (which can't do special functions) and 1 fat one (which can do special functions) as a single stream processor.
Nvidia counts a cluster of one normal processor and one special function processor as a single stream processor.
- note that neither processor is comparable to its competitor's version.
So the total number of individual processors in:
AMD's Radeon HD 3870 - 320 SP running at 775 MHz - Total GigaFLOPS = 496. Core clock speed = 775 MHz. Mem speed = 2250 MT/s GDDR4.
Nvidia's GeForce 9800 GTX - 256 SP running at 1125 MHz - Total GigaFLOPS = 416. Core clock speed = 650 MHz. Mem speed = 970 MT/s GDDR3.
- Reference specifications. The core clock speed is what the rest of the chip (i.e texture units, command processor, output units etc.) runs at. Since both cards I've mentioned have identical ammounts of RAM and an equally wide bus to RAM, their RAM bandwidth is directly related to the speed of their RAM. GDDR4 has higher latency than GDDR3 so although it runs faster it has to do this to 'catch up', it offsets this by being able to run at higher speeds.
But the real reason for the Radeons falling behind the GeForce is the difference in texturing capacity: 12.4 GigaTexels verses 41.6 GigaTexels. I'm sure that the fact Nvidia works with almost every game developer has nothing to do with. .