Set apart the 5% better IPC, what is the improvement scale of the two other metrics, perf/watt and perf/$..?.
Do you have numbers, or even raw estimations of your own, that would suggest significant progress were made on thoses two parameters.?.
It is true, Broadwell makes no sense provided you aren't interested in improved performance (performance/clock, performance/watt, performance/dollar, etc.).
Provided none of those metrics are relevant or of interest to you (the royal you), you are better off waiting for something that won't be released until a later date.
Set apart the 5% better IPC, what is the improvement scale of the two other metrics, perf/watt and perf/$..?.
Do you have numbers, or even raw estimations of your own, that would suggest significant progress were made on thoses two parameters.?.
Last I saw, the top Broadwell CPU was 65W (which is ~30% more perf/watt over IVB, the last tick). It's just that Skylake-K will have a 95W TDP and likely be a better choice for an enthusiast (but sadly will be making a pretty late appearance it seems).
Set apart the 5% better IPC, what is the improvement scale of the two other metrics, perf/watt and perf/$..?.
Do you have numbers, or even raw estimations of your own, that would suggest significant progress were made on thoses two parameters.?.
The rate of improvement isn't too bad, from the perspective of a user with Ivy Bridge. It works out to be roughly 17% YoY for the CPU, comparing GT2 to GT2. It's also about 52.5% faster, comparing my 3317U to the 5300U, stock to stock.
And Core M processor with Intel 14nm process belonging to, Broadwell-U series processors. will be divided into "2 + 2" and "2 + 3" versions.
Yeah, I think you're right about 4200U being the 3317's sucessor. Haswell was really nice from a battery life standpoint though, and we should see further gains with Broadwell, in addition to performance. Graphics should be nice, too (relative to Haswell), although the low clock speed is puzzling.It's not that much faster actually, because we had a literal standstill in the Haswell generation.
The 3317U gets 2.41 points in Cinebench R11.5, and probably runs at 2.4GHz to achieve that. I'd say 3317U's successor was really 4200U, and 4200U's successor 5200U, but let's see 5300U as well since you mentioned it.
Cinebench score to clocks ratio:
-Ivy Bridge 1:1
-Haswell: 1.1:1
-Broadwell: 1.15(?):1
3317U: 2.4 points = 2.4GHz
5200U: 2.5GHz = 2.88 points
5300U: 2.7GHz = 3.1 points
That's probably an optimistic scenario. I think you can expect 25% going to 5300U and 20% to 5200U. Still not bad since we got essentially nothing from a "Tock" Haswell. And at least confirms Broadwell won't be a disappointment performance-wise unlike Haswell.
Another good news is that Iris parts are available at 15W. On Haswell it took 28W to do so. Also the clock frequency is rather high. No wonder there's no M anymore. The U's replace them completely.
The fastest Haswell-U parts at launch were:
Core i7 4500U - 1.8GHz (3GHz Turbo) 2C/4T with GT2 graphics
Core i7 4650U - 1.7GHz (3.3GHz Turbo) 2C/4T with GT3 (no eDRAM) @ 200/1100 MHz
Now we have:
Core i7 5600U - 2.6GHz (3.2GHz Turbo) 2C/4T with GT2 graphics
Core i7 5557U - 3.1GHz (3.4GHz Turbo) 2C/4T with GT3e (eDRAM) @ 300/1100MHz (50% higher base clocks)
That's a massive CPU and graphics performance increase for a Tick(+), kinda reminds me of the Sandy Bridge ULV -> Ivy Bridge ULV transition (but better).
It's not that much faster actually, because we had a literal standstill in the Haswell generation.
The 3317U gets 2.41 points in Cinebench R11.5, and probably runs at 2.4GHz to achieve that. I'd say 3317U's successor was really 4200U, and 4200U's successor 5200U, but let's see 5300U as well since you mentioned it.
Cinebench score to clocks ratio:
-Ivy Bridge 1:1
-Haswell: 1.1:1
-Broadwell: 1.15(?):1
3317U: 2.4 points = 2.4GHz
5200U: 2.5GHz = 2.88 points
5300U: 2.7GHz = 3.1 points
That's probably an optimistic scenario. I think you can expect 25% going to 5300U and 20% to 5200U. Still not bad since we got essentially nothing from a "Tock" Haswell. And at least confirms Broadwell won't be a disappointment performance-wise unlike Haswell.
Another good news is that Iris parts are available at 15W. On Haswell it took 28W to do so. Also the clock frequency is rather high. No wonder there's no M anymore. The U's replace them completely.
5557U with HD graphics 6100 shouldn't have edram and I somehow doubt this is a 15W part.
DiskId = "Intel(R) Graphics Media Accelerator Driver for Windows"
Intel = "Intel Corporation"
; HSW Classic
iHSWGT1D = "Intel(R) HD Graphics"
iHSWGT1M = "Intel(R) HD Graphics"
iHSWGT15D = "Intel(R) HD Graphics 4400"
iHSWGT2D = "Intel(R) HD Graphics 4600"
iHSWGT2M = "Intel(R) HD Graphics 4600"
; HSW ULT
iHSWGT1UT = "Intel(R) HD Graphics"
iHSWGT2UT = "Intel(R) HD Graphics Family"
iHSWGT3UT = "Intel(R) HD Graphics 5000"
iHSWGT3UT28W = "Intel(R) Iris(TM) Graphics 5100"
iHSWGT2UX = "Intel(R) HD Graphics Family"
iHSWGT1ULX = "Intel(R) HD Graphics"
; HSW CRW
iHSWGT3CW = "Intel(R) Iris(TM) Pro Graphics 5200"
iHSWGT3CWDT = "Intel(R) Iris(TM) Pro Graphics 5200"
; HSW SRVR
iHSWSVGT2 = "Intel(R) HD Graphics P4600/P4700"
; BDW HW
iBDWULXGT2 = "Intel(R) HD Graphics 5300"
hopefully we can see a 25% increase. I'd hope we see better than 2.88 pts though, hopefully we can see around 3 pts. with the 5200U.
3317U came at Ivy Bridge's introduction. They released 3337U with 100MHz higher clocks later. And it wasn't a business oriented part like the 4300U. That practically puts it on par with 4200U. Ok, maybe you didn't know and I didn't mention that.The 4200U still achieved 2.5 pts, so not exactly a standstill, but a 5% improvement is nothing to cheer about.
That's still very good, if they moved from requiring 28W for 2+3 part in Haswell to 15W 2+3 part, and 28W 2+3e part. I mean, the CPU is faster too. You are talking 3.9 points on a dual core part. That's why the -M series SKUs are gone. I figure though it might be a typo and 2+3 parts are really 28W and 2+3e parts are 35/45W parts only for desktops and special mobiles.Perhaps that's a little too optimistic.
The biggest reason not having the eDRAM on the 15W and lower parts is because the current eDRAM modules would take up lot of TDP. 3.5-4.5W for the eDRAM on current Crystalwell parts are massive 1/3rd of the TDP level, while is relatively insignificant for a 28W part. The eDRAM presentation this year reduces standby power to 1/5-1/4th but they said nothing about peak power.15W TDP ''U'' chips with Iris Pro should arrive with Skylake.
The GPU? If you look at benches in Notebookcheck, the HD 4400 chips really aren't running higher than 1GHz in any meaningful game. Now if Intel engineers figure out that setting it to 1GHz benefits it in some way that's good enough for me.although the low clock speed is puzzling.
Hey guys look at the Iris 6100 chart:
TDP: 4MB
Tj(deg): 15W
Intel SBA: 105
Perhaps we should take that to mean 15W is a typo.