I think NV has to use ultra high leakage chips to reach GTX480 clock.
NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
Why would there be a correlation b/w high leakage and high clocks?
High leakage means chip can achieve certain clock at lower voltage.You cant bump up voltage without limit,so high leakage is good for clock.
For example,AMD TWKR is high leakage chip.
I was blasting you about using metal spins to reduce leakage. Linking I/O performance and interference to leakage isn't much better...
You're good about knowing dates and early specs, much better than most. Don't spoil it by overreaching.
Each 64 bit memory controller in GF100 is tightly coupled to an octo-ROP and 128 kiB L2-Cache, disabling one means disabling the other stuff top. That is, unless Nvidia lied.
Now, GT215 has eight ROPs and a two-channel interface to memory, no word on how large and of what type the L2 is. So that means, they must have greatly reworked the memory controller at least in order to be able to service twice as many ROPs. I can hardly imagine, that they've left any potential weaknesses untouched in the process.
Now, GT215 has eight ROPs and a two-channel interface to memory, no word on how large and of what type the L2 is. So that means, they must have greatly reworked the memory controller at least in order to be able to service twice as many ROPs. I can hardly imagine, that they've left any potential weaknesses untouched in the process.
High leakage => high temp => >hysteresis than expected?
Ah thanks, that clears things up. Maybe I end up translating things wrong, could replace leakage with "interference" or things like that.
I just noticed that the magic 256 driver won't be the performance booster everyone is hoping for, but rather the first one to officially support 3D Vision Stereo Surround.
And re: die size.. yeah.. for every good cookie there's a bad one. 529mm2
and re: high leakage parts. all the factory OC and 5970 Cypress dies are also high-leakage parts.
But max clock is determined by the max clock achievable in the "worst" part of the chip. Whereas the high leakage would be a result of the leakage of all transistors. Is there really any correlation between the average and the extremes?
Anyone measured voltages on GTX470? ht4u only had a GTX480, and I don't think I've seen voltage measurements anywhere else.
The products may be bounded differently though. At 225W the lower leakage range may be reserved for the 470, while with a higher TDP on the 480 the higher leakage parts may be reserved for that.
Its just something you you often find a correllation between.
Does anyone know where the post in question is located? I have been unable to find it and am currently involved in a discussion about it elsewhere, I'd like to cite it if at all possible.
Not quite what I was thinking of. ISTR discussion in this thread about GF100 consuming less power when running at lower temperatures. This stems from power consumption results varying between review sites, with some sites using higher fan speeds and achieving seemingly lower power consumption.
That's inherently physics, cool conducts better than warm, right?
Mintmaster
Veteran
Semiconductor physics failResistance to electron flow most certainly increases with heat.
Er, well, while it is true that increases in temperature increase the mobile electrons/holes in semiconductors, processors make use of doping for the majority of their mobile electrons/holes, so that effect is small, and is generally offset by other deleterious effects, as far as I am aware.Semiconductor physics fail
Similar threads
- Deleted member 2197
- Architecture and Products
