Via C7 announced
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Via C7 announced
http://www.via.com.tw/en/products/processors/c7/
An competitor for Pentium M by the looks of things. Hopefully Via will release mini-itx boards using this chip and it will be interesting to see how performance and pricing pans out.
An competitor for Pentium M by the looks of things. Hopefully Via will release mini-itx boards using this chip and it will be interesting to see how performance and pricing pans out.
I like the way they show performance as "score per watt". Extrapolating from the comparisons with the P4M with its almost double TDP, it looks like the C7 will perform, clock-for-clock maybe half as fast as the P4M. Which is an improvement, to be sure, but nothing on the surface to get too excited about.
No info on socket compatibility either.
No info on socket compatibility either.
Last edited by Rusty075 on Fri May 27, 2005 7:13 am, edited 1 time in total.
Hmm. The image linked below would seem to indicate that performance is compared with the Pentium M not P4M:Rusty075 wrote:I like the way they show performance as "score per watt". Extrapolating from the comparisons with the P4M with its almost double TDP, it looks like the C7 will perform, clock-for-clock maybe half as fast as the P4M. Which is an improvement, to be sure, but nothing on the surface to get too excited about.
No info on socket compatibility either.
As the Pentium M is a low power chip too this would indicate much better performance than if it was compared to P4M. Is this in comparison to Pentium M as is indicated or is it just a bit of disingenuous marketing?
The little I'd heard about the C7 indicated it would be quite a high-performing chip.
As for the socket, I'd guess it won't be compatible with any of the current standards.
-
- *Lifetime Patron*
- Posts: 301
- Joined: Sat Dec 14, 2002 11:29 pm
- Location: Dublin, CA / Liverpool UK
I love that performance per watt graph too... it makes me want to see the details of that: as in which Pentium M did they use and what TDP figures did they use for it, as well as what the actual scores were [and why did they "normalize" to a 1.5ghz C7 and not use one of the other speed grades].
speaking strictly in terms of performance:
If Rusty's estimate is correct (VIA @ 2ghz is about the same as a 1ghz P-M), then that suggests the actual chip is pretty much the same architecture as Via's current cpus, with some generational improvement. They may have improved their assisted mpeg playback, which would be a good thing.
speaking on price:
Via's prices have always been affordable so if anything, that's where they ought to get some business.
Their claim of an average of 1watt suggests Via's real big change is an amazingly low idle power setting! I do see where they say they use .1 watt at idle... but that's so amazing I'm thinking more info is needed
speaking strictly in terms of performance:
If Rusty's estimate is correct (VIA @ 2ghz is about the same as a 1ghz P-M), then that suggests the actual chip is pretty much the same architecture as Via's current cpus, with some generational improvement. They may have improved their assisted mpeg playback, which would be a good thing.
speaking on price:
Via's prices have always been affordable so if anything, that's where they ought to get some business.
Their claim of an average of 1watt suggests Via's real big change is an amazingly low idle power setting! I do see where they say they use .1 watt at idle... but that's so amazing I'm thinking more info is needed
-
- *Lifetime Patron*
- Posts: 301
- Joined: Sat Dec 14, 2002 11:29 pm
- Location: Dublin, CA / Liverpool UK
Mariner,
I assumed Rusty meant the P-M. And the performance-per-watt graph correlates to previous Via graphs on the subject (if memory serves me right) which suggests the performance of the C7 is not that different from their current C3 line.
As far as hearing the C7 is a high-performance chip, I haven't heard that except relative to the previous generation of Via chips (improved FPU and SSE3 support the big improvements).
I'm guessing we're still talking about a cpu that clock-for-clock will not match the performance of the P-M,Athlon 64, or Athlon XP.
Price per performance & or performance per watt are still the tools that Via uses to sell these cpus, which suggests that a straight performance comparison would not put the c7 in a good light.
ps. I *like* via chips. I have so far resisted the urge to tinker with them though...
I assumed Rusty meant the P-M. And the performance-per-watt graph correlates to previous Via graphs on the subject (if memory serves me right) which suggests the performance of the C7 is not that different from their current C3 line.
As far as hearing the C7 is a high-performance chip, I haven't heard that except relative to the previous generation of Via chips (improved FPU and SSE3 support the big improvements).
I'm guessing we're still talking about a cpu that clock-for-clock will not match the performance of the P-M,Athlon 64, or Athlon XP.
Price per performance & or performance per watt are still the tools that Via uses to sell these cpus, which suggests that a straight performance comparison would not put the c7 in a good light.
ps. I *like* via chips. I have so far resisted the urge to tinker with them though...
-
- *Lifetime Patron*
- Posts: 301
- Joined: Sat Dec 14, 2002 11:29 pm
- Location: Dublin, CA / Liverpool UK
ack, another post in a row from me:
http://mini-itx.com
has the news this on their front page and this snippet sheds more light on power consumption:
Clock speeds are expected to reach up to 2Ghz, with power consumption varying between 0.1W at idle and 20 Watts peak on a 2Ghz chip, with bus speeds up to 800Mhz possible. The C7's tiny 30 mm square die includes technology to reduce processor speed when computing loads are light: at 1GHz, consumption is a maximum of 3 Watts, and at 1.6GHz it is 12 Watts. This bodes well for low temperatures - though the fastest variants may still require some active cooling.
http://mini-itx.com
has the news this on their front page and this snippet sheds more light on power consumption:
Clock speeds are expected to reach up to 2Ghz, with power consumption varying between 0.1W at idle and 20 Watts peak on a 2Ghz chip, with bus speeds up to 800Mhz possible. The C7's tiny 30 mm square die includes technology to reduce processor speed when computing loads are light: at 1GHz, consumption is a maximum of 3 Watts, and at 1.6GHz it is 12 Watts. This bodes well for low temperatures - though the fastest variants may still require some active cooling.
Yeah, I did mean the P-M, as Via lists on their graph. Got befuddled by the jargon, again. (actually the C7 would look even better compared against the P4M)
Based on what we know so far it looks like an incremental change, coupled with a die shrink and SOI, as opposed to a ground-up new CPU. Sounds like their adding a version of CnQ/Steedstep to drop the idle wattage down even lower.
Lower wattage is always a good thing, and with the right mobo chipset these could be plenty powerful for what they appear to be aiming them at.
Based on what we know so far it looks like an incremental change, coupled with a die shrink and SOI, as opposed to a ground-up new CPU. Sounds like their adding a version of CnQ/Steedstep to drop the idle wattage down even lower.
Lower wattage is always a good thing, and with the right mobo chipset these could be plenty powerful for what they appear to be aiming them at.
'Have to wonder whether the tiny die size might create some problems.
If the proportions in the above image are correct, then the C7 die would have approximately 1/3 the surface area of the C3 die, which means that there would only be 1/3 the contact area with the heatsink. The heatsink contact material would therefore have to have 3 times the thermal conductivity to be as effective. (On the other hand, with the necessary contact area being so small, maybe you could coat it with diamond. )
The accuracy of the figures in the image is in serious doubt. The text on the Via page says that the C7 die is 30mm2. The actual size of the C3 die is 25mm2!
You can assume that the Via engineers got it all figured out, but I wonder how well the situation bodes for fanless cooling.
If the proportions in the above image are correct, then the C7 die would have approximately 1/3 the surface area of the C3 die, which means that there would only be 1/3 the contact area with the heatsink. The heatsink contact material would therefore have to have 3 times the thermal conductivity to be as effective. (On the other hand, with the necessary contact area being so small, maybe you could coat it with diamond. )
The accuracy of the figures in the image is in serious doubt. The text on the Via page says that the C7 die is 30mm2. The actual size of the C3 die is 25mm2!
You can assume that the Via engineers got it all figured out, but I wonder how well the situation bodes for fanless cooling.
Good god. The yields on it will be incredible, I'm sure it'll be very cheap to make.Reachable wrote: The accuracy of the figures in the image is in serious doubt. The text on the Via page says that the C7 die is 30mm2. The actual size of the C3 die is 25mm2!
I wouldn't be too worried about the cooling, even with the smaller die, the power density will still be much lower than a lot of the P4s and Athlon XPs.
-
- Posts: 1608
- Joined: Tue Jan 04, 2005 4:02 pm
- Location: United States