Nv30 Clock rates?
- Thread starter sancheuz
- Start date
Humus said:
Because that's the most misused term in the industry. Bandwidth refers to the width of the band (Hz), but makes no reference to the band *depth* (bits per sample). The correct term is throughput (bandwidth*band depth=bits/sec), but, for whatever reason, the computer industry has veered from the correct engineering terms and decided that "bandwidth" = throughput.
I say we just state the throughput.
Mize
Gah! The Inquirer once again reports incorrectly old news ...
http://www.theinquirer.net/?article=5665
They make the same mistake of thinking the speeds shown in the picture are actual MHz ratings, and not kHz ratings.
I'm surprised it's taken them this long to report this falacy.
--|BRiT|
http://www.theinquirer.net/?article=5665
They make the same mistake of thinking the speeds shown in the picture are actual MHz ratings, and not kHz ratings.
I'm surprised it's taken them this long to report this falacy.
--|BRiT|
I remember from statistics the saying that if you took monkeys and locked them up long enough with a typewriter they would write the complete works of Shakespeare.
Somehow I think the Inquirer really liked this idea and their executives are just trying to compress the timescales, but stuck with the monkeys and the typewriters
Inquirer / used toilet paper / what's the difference?
Somehow I think the Inquirer really liked this idea and their executives are just trying to compress the timescales, but stuck with the monkeys and the typewriters
Inquirer / used toilet paper / what's the difference?
Mize said:Humus said:
Because that's the most misused term in the industry. Bandwidth refers to the width of the band (Hz), but makes no reference to the band *depth* (bits per sample). The correct term is throughput (bandwidth*band depth=bits/sec), but, for whatever reason, the computer industry has veered from the correct engineering terms and decided that "bandwidth" = throughput.
I say we just state the throughput.
Mize
Hmm...bandwidth in my training is not related at all to digital signal transfer, atleast fixed frequency bit transfers. For instance, it would be a factor in any case where frequency modulation or frequency multiplexing take place (and, as a result, the range of frequencies used has some relevance to the amount of information transferred). So, yes it is the width of the band, but that still has nothing to do with (my understanding of) DDR for example, since the frequency isn't a "width" it is a clock.
As we progressed frequency modulation as used in data transfer has been a common theme, and the concept of bandwidth has been distorted (again, as far as I understand) to be semi-synonymous with amount of data transferred, even though in this case it has no technical relevance to the origin of the term.
Only the bitrate and cycle measurment terms make sense to me in an absolute sense. I think part of the familiarity of the usage is because actual "bandwidth" in many cases controls bitrate (it should, for example, work that way for DSL and cablemodems, and wireless communications standards).
Then again, maybe this distortion of the original term has become official in some way...I wouldn't be familiar enough with memory technology to say.
On a side note here we do have affirmation from ATIs CC that rumored clock speeds (eg 400Mhz?) of ATIs next gen product (R350 + RV350?) are truthful. Now whether or not it is a .15um or .13um product they were referring to is another matter as ATi intends to have both .15um and .13um contributions at that time.
I don't know what ATI have said, but someone sent me a mail yesterday with some voltage mods of Radeon 9700 - some of these were hitting 400-450MHz.
Doomtrooper
Veteran
Pin 18 on The SC chip
3300 ohms = 1.7volts from 1.51 default
Using a 5k vr 4500 about 1.62
3300 ohms = 1.7volts from 1.51 default
Using a 5k vr 4500 about 1.62
DaveBaumann said:
Yeah I have got wind of the higher 450MHz but that is one hell of a difference from 325MHz. It almost seems unrealistic. I am thinking that the 400MHz "engineering sample" that AnandTech got a hold of a while back would be a more reasonable item. But who knows..... apparently ATi has been able to squeeze more out of the .15um process then they had originally thought. Anyhow this is a bit OT.
Just thought that it was an interesting "sidenote".450MHz just by bumping the voltage, now that is truly impressive. Seems ATI did an excellent job architecting the R300!
Seems to re-affirm my view that if ATI launches a product around christmas to "counter" the NV30, it will definitely not be anything more than a speed-bumped R300 with faster memory.
However, this report of 450MHz, was it achieved with air cooling? If it's water (and possibly a TEC), it isn't really as impressive.
Details, please! Including benchmarks, heheh!
Anyway, people complain 3Dmark is too CPU limited. I'd like to see GF4 and R300 comparisons at 1600 resolution, and/or with FSAA. That puts more strain on the video card. Most run 3dmark at the standard 1024 res and of course that makes the R300 CPU limited on just about every system out there... I still hate 3dmark as a benchmark, don't get me wrong, but it would still have some curiosity value...
*G*
Seems to re-affirm my view that if ATI launches a product around christmas to "counter" the NV30, it will definitely not be anything more than a speed-bumped R300 with faster memory.
However, this report of 450MHz, was it achieved with air cooling? If it's water (and possibly a TEC), it isn't really as impressive.
Details, please! Including benchmarks, heheh!
Anyway, people complain 3Dmark is too CPU limited. I'd like to see GF4 and R300 comparisons at 1600 resolution, and/or with FSAA. That puts more strain on the video card. Most run 3dmark at the standard 1024 res and of course that makes the R300 CPU limited on just about every system out there... I still hate 3dmark as a benchmark, don't get me wrong, but it would still have some curiosity value...
*G*
