Separate names with a comma.
Discussion in 'CPUs and Overclocking' started by Sweepr, Apr 16, 2015.
Well, I dont know, do I, because no 14nm APUs exist, and none of them have HBM. My grandson is graduating high school this month, so I can hardly wait till 2017 for this theoretical APU. By the way I do have an AMD dgpu in my gaming desktop, so it is not like I am out to get AMD. If they build the product, and it fits my needs, I will buy it. To be honest, if there is one thing that will prevent me from buying an AMD product, it is the continual hype promoted on these forums. That and personal insinuations that I am so biased I would not buy an AMD product.
But again, this is what mystifies me about intel's igpu lineup. Why put iris pro on a 15 watt chip? The normal igp is good enough for everyday use, and you cant tell me one is going to be doing much gaming or other gpu intensive work in a 15 watt TDP.
Simple: customers (i.e. OEMs) asked Intel for it.
In particular, I'd wager Apple asked Intel for it.
Keep in mind 15W TDP chips can also operate at 25W TDP. (ex: 4510U operates with a base speed of 2.6Ghz at 25W TDP)
Ah! If that extra power draw also applies to improved iGPU speed, then it makes allot more sense. More demanding GFX (especially short term) can make use of higher performance. The downside is that fanless designs would be limited in long term performance due to throttling under a persistent temperature excursion (say, like running a game). It's really amazing how many people play games, even if most are much less demanding than a AAA PC title.
Hey all, when do you guys think we will have 100-chipsets available? Some rumors point out they will be shipping/available this month, any news on that? besides the iGame one.
When the CPUs ship?
I guess we start see mobos selling in july. But with lots of mobo leaks startiing in june.
At the same time as when the CPU's go on sale. It wouldn't be a very good idea to buy a motherboard that you can't use. Before reviewers get their hands on it etc.
Mobile Skylake GT4e might come earlier than expected (but surely not before Q1 2016). 3dmark11, Vantage projections over BDW GT3e up to 50%. We don't have BDW GT3e 47W scores unfortunately.
"4K ISP" -- they integrated an image signal processor onto the Skylake Core M chips? Sensor hub, too?
Neat. Core M is finally becoming a proper tablet SoC. Next up is probably the PCH for the -Y and -U parts come Cannonlake.
Actually, I am not that impressed with "only" a 50% improvement. The reason is, you have 33% more EUs for only a 50% improvement. And that improvement could be just from lower power use allowing the igp to run at higher turbo. Seems like Gen 9 is not a quantum improvement in architecture either. OTOH, if BW is a 20 or 30% improvement over Haswell and Skylake brings another 50%, we could have a decent low/mid range gaming capability without a discrete card. Question is, what will be the availability and price.
Wow, nice find once again mikk.
- Skylake-Y draws up to to 60% less power (SoC only) compared to current Broadwell-Y (Core M). Nice gain considering they are both 14nm products.
Just like Ivy Bridge ULV vs Sandy Bridge ULV and the first ultrabooks I feel like Broadwell-Y was a well executed first attempt, but Skylake-Y might be faster, more efficient, more integrated and a lot more popular among premium fanless tablets and convertibles.
- Not sure about Vantage but dual-channel HD5500 is close to 30% faster than dual-channel HD4400 @ 3DMark 11.
Assuming something similar for Broadwell GT3e vs Haswell GT3e - Skylake GT4e should provide 2x Haswell GT3e performance in this benchmark. Benchmark scores were never a big problem for previous Intel iGPUs, hopefully they solved some of the architectural bottlenecks and there will be bigger gains in actual games (especially with MSAA and filters on).
- I'm more confident that they will eventually launch a desktop Skylake 4C+GT4e sometime in H1-2016 now.
I would expect such a structural big change from the Oregon core, means Kabylake/Icelake might come as a full SoC for the U/Y lineup.
There was never a linear scaling from an EU increase. A 50% EU addition wouldn't result in a 50% fps improvement across-the-board. There are other bottlenecks such as bandwidth or texturing. In 3dmark11 Gen8 looks quite good already, in games not so good. Although this is based on ULV Gen8 reviews.
People really need to read the fine print.
I read that, that's why I said up to, not on average.
There's clearly room for improvement when it comes to VP battery life and it's nice to see Skylake-Y improves this:
This points to a much enhanced video decode block in Gen. 9 over Gen. 8. Makes sense as all of the big media improvements were set to come in Gen. 9.
That makes sense, actually -- wait for the next "Tock" for such a major platform-level change esp. since I'd assume OEMs want Cannonlake to be socket-compatible with Skylake.
BDW-Y/U was also 60% lower power, and HSW probably similar, so nice momentum.
I wasn't downplaying the benefits, just making it clear we were talking about video decode.
On the plus side, the +35% playback time was noted versus an efficient CoreM tablet (11.6" screen, LPDDR3).
But is Iris pro all about gaming performance? I though I read somewhere that having 64-128 MB of cache lets them shut down external memory/buses to a larger degree, meaning that it is a power saving trick?
How well do you guys think mobile GT4e will do in comparison to a given AMD/Nvidia video card?
The Broadwell C should give some vague idea whenever it finally appears....
It might be starting to get there, although it'll likely spend a fair bit of its lifetime vs the next gen 14nm/HMB stuff from the GPU people which should lift the bar a long way.
Although Intel will then likely get the odd die shrink in before the GPU folk move on again etc
I don't think so.