Thanks for the link on the resonant clock mesh. I think I understand the 5% for application and the 10% reduction at idle. What is the -25% "clock power" reference about?
I may have missed something reading this (thanks, btw, very interesting) was weather or not the cleaner clock signals prevailed above Fmax. While power saving drop off I wonder if increasing clock driver strength to RM above 4GHz would give a clean signal and enable higher clocks to be reached (for enthusiasts)?
Now all we need are fatter cores, more execution ports, faster decoding, a larger L1 and smaller faster L2 and we'll be getting somewhere
What article states(what you quoted) is the opinion of the author of the article. What AMD stated was "they stay committed to performance processor segment"(that is the exact one sentence reply they got,nothing more and nothing less). There is no double wording there.They did not deny nor confirm. As your article states
So lets wait and see.
The "clock power" is the power consumption by the clock driver.
For example, say the CPU uses 100W. Of that 100W, 20W of it is from the clock power. The other 80W is for the cores, the cache, and the NB.
Reduce the clock power by 25%, from 20W to 15W, and you saved 5W.
5W reduction means 5% reduction for the overall CPU's power usage. It now uses 95W instead of 100W.
The "clock power" is the power consumption by the clock driver.
For example, say the CPU uses 100W. Of that 100W, 20W of it is from the clock power. The other 80W is for the cores, the cache, and the NB.
Reduce the clock power by 25%, from 20W to 15W, and you saved 5W.
5W reduction means 5% reduction for the overall CPU's power usage. It now uses 95W instead of 100W.
Resonant mesh degrades the clock signal, it does not improve it. RM skew was 7.2ps versus 6.5ps for the conventional clock. (see page 29)
They gave some priority towards minimizing the degradation (page 28). In the end they optimized the mesh such that it only degrades fmax by 0.2% (page 33).
It is a novel technique but it isn't without its drawbacks. But this is just a first-round implementation. Maybe there is a lot more that can be done going forward as AMD refines the implementation.
I'm not buying this rumor, AMD FX Zambezi CPU's have been on a steady price decrease ever since release to this day, which usually indicates another impending SKU launch.
I think it will have the same one core/half core turbo like 8150 (4.2Ghz max). That doesn't leave much room for full core turbo though,the only option is the 4.1GHz (4-4.1-4.2) which is kinda pointless since it is so small that it would be unnoticeable. Maybe Turbo 3.0 spec means that 8350 can run with all cores under full load @ 4.2Ghz but that's unlikely IMO. In any case, 8350 should at least clock a bit better and be more efficient(IPC) than the 8150 and for the 10% higher price it should be decent chip from price/perf. POV.
The relatively small frequency increase could either mean that they have concetrated on improving the per-clock performance, rather than the clock rate, or, they have barely managed to improve it at all.
I do hope it's the former - I need a replacement for my 960t.
I'm not buying this rumor, AMD FX Zambezi CPU's have been on a steady price decrease ever since release to this day, which usually indicates another impending SKU launch.
I think it will have the same one core/half core turbo like 8150 (4.2Ghz max). That doesn't leave much room for full core turbo though,the only option is the 4.1GHz (4-4.1-4.2) which is kinda pointless since it is so small that it would be unnoticeable.
aMd needs at least 6Ghz to compete with an i7 920. aMd is done. They are trash, garbage, finished. Low power, low cost, low performance, low low low low...and they are high to think they have a chance at competing with Intel.
Ya, the price cuts are coming.
I don't think anyone here wants $830+ 8800GTX and $650/$500 GTX280/GTX260 days back, or a Core i5 processor for $400.
aMd needs at least 6Ghz to compete with an i7 920. aMd is done. They are trash, garbage, finished. Low power, low cost, low performance, low low low low...and they are high to think they have a chance at competing with Intel.
Ya, the price cuts are coming.
Source
This rumour doesn't seem reasonable.
July 2012 leaked AMD slide, published today
The eight-core FX-8350 is confirmed to ship with 4.00 GHz nominal/base clock speed, with 4.20 GHz TurboCore speed. The six-core FX-6300 ships with 3.50 GHz nominal, and 4.10 GHz TurboCore speed. The quad-core FX-4320, on the other hand, ships with the same clock speeds as the FX-8350.
You need to calm down man. If you hate AMD, keep that to yourself. We need competition in the CPU and GPU space. You can enjoy your Intel/NV love obsession privately in a more mature manner. I don't think anyone here wants $830+ 8800GTX and $650/$500 GTX280/GTX260 days back, or a Core i5 processor for $400. No one is expecting Piledriver or Steamroller to actually beat Haswell but as long as AMD continues to improve, it's something interesting for us to read about. Also, not a single company in the world has been able to successfully compete long-term with Intel because on top of their excellent engineering, they are a step ahead in the manufacturing node process. That almost guarantees that AMD either has to undercut Intel or engineer a more efficient processor. That's very difficult to do since it's not like Intel hires University of Hawaii grads. Regardless, any progress on the AMD side is better than nothing at all.
In case you need to be reminded of what happened to CPU prices when Intel was behind during Pentium-D days:
I cannot believe this is a serious post.
anyways as Russiansensation wrote, if AMD don't try then we're screwed as shown by the prices of CPUs when Intel were behind.
You need to calm down man. If you hate AMD, keep that to yourself. We need competition in the CPU and GPU space.