maybe because, who buys a llano cares about igpu, and the ones who buy a Ivy don't care about igpu
I wouldn't let an es chips results bother me. Its funny with llano anyone who got their hands on one really discarded the cpu performance and went to gpu performance and O/Cs .
With IVB we already know the performance will be better depending on app. 5%-25% better / But none who have O/Ced IVY have tried O/Cing the iGPU. Odd that!
maybe because, who buys a llano cares about igpu, and the ones who buy a Ivy don't care about igpu
With all due respect, you are dead wrong. ^_^No, I need moar corez!!
Well I'm assuming that in 1-2 years more software will be using all cores it can see.
I'm not expecting too much with these instruction extensions, they are mostly limited to specific situations. AVX helps SB a lot in a few cases, but hardly a game changer compared to predecessor. I could be wrong, but I'm skeptical that AVX2 will bring much more.
cheater.
work for Intel=discounts
I'm going to get the 3770k but not because of performance needs over my 2600k; rather, because I'm really curious to play around with an IC that has new 22nm 3D xtors under the hood.
That seems 99% impossible. If Intel did that there would be very little to no reason to get SB-E, even the six-cores. Given what we've seen historically play out, there's some things that stand out:
1. Ticks are small architectural revisions that introduce a new process node. By virtue of having a smaller die on an already existing architecture, power consumption decreases. This also allows for small revisions that couldn't make it before to be implemented. These revisions mean a small jump in IPC, for Intel typically 5% as we've seen going from Conroe to Penryn and from Sandy Bridge to Ivy Bridge. Having a smaller die and lower power consumption typically also means lower temperatures, which enables higher clock speeds. If you have high leakage or high temperature at normal voltage with a new process, however, clock speeds may improve little if any.
2. Tocks are bigger architectural revisions that bring heavier changes, whether it be to the cache, IMC, or the execution units themselves (latency, branch prediction, etc). These result in bigger changes when it comes to IPC, and typically increase clock speeds forward. Going from Penryn to Nehalem we saw a ~17% increase in IPC, and going from Westmere to Sandy Bridge we saw an ~11% increase in IPC.
Now let's assume that Haswell has 10% higher IPC than Ivy Bridge, which has 5% higher IPC than Sandy Bridge. Let's also assume that Ivy Bridge, due to what is already being reported, improves clock speed over Sandy Bridge by 0-5%. Let's assume the situation regarding clock speeds doesn't get better for Intel by the time Haswell comes out because of their quest for efficiency.
Sandy Bridge-E uses the Sandy Bridge architecture, and it's a safe bet given history that Haswell will have around 15% higher IPC than Sandy Bridge. Let's assume the clock speed situation improves little and you can only achieve 5% higher clock speeds on Haswell compared to Sandy Bridge.
This leaves us with a 20% improvement over Sandy Bridge. Six-Core SB-E has a theoretical 50% more computing power than Quad-Core Sandy Bridge, but because of lower core scaling as you add cores even in the most multi-threaded scenarios that becomes 40-45% and in most multi-threaded programs will mean a difference of around 30%.
Now, why would anyone in their right mind pay over twice for a Sandy Bridge-E processor when Haswell will only be 10-20% slower in multi-threaded scenarios, even though it has 50% less cores and much lower power consumption and much lower platform costs? Not gonna happen. Ivy Bridge-E is definitely gonna be released.
Next, AVX2 is a far bigger deal than AVX. For the first time in the history of mainstream processors, every (relevant) scalar instruction will have an SIMD vector equivalent! This is fundamental because it makes it possible to vectorize most performance-critical code loops.
Note also that loop vectorization is an optimization that can be handled automatically by compilers, and will be supported by all major ones before Haswell hits the street. Furthermore, throughput computing frameworks like OpenCL can seamlessly support GPUs and CPUs (with or without AVX2) so application developers don't even have to explicitly target AVX2 to get a speedup.
Not a discount. This is my work computer and they would like me to have a new system with each generation so as to make sure I have experience with them when talking to you guys. So I already have ordered all the parts and am just waiting to get them and put it together.
At home I have an Intel® Core i7-860 that I would love to upgrade but right now the budget just wont allow me to do that.
i don't know that this is true often enough to say 'typically.' temperature is a function of power per unit surface area and cooling capacity. if your power consumption decreases by 20% but your surface area decreases 50% from the shrink, your power per unit area has increased markedly. in that case, any reduction in temperature is a result of increased cooling capacity. cooling capacity has increased a ton over the last 15 years. we went from small glue on heatsinks with no particular direct airflow needed to needing large amounts of airflow to keep the same temps, and needing very large heatsinks to do it while maintaining quiet.Having a smaller die and lower power consumption typically also means lower temperatures, which enables higher clock speeds.
I think I will start off with a 3570k since I only single GPU game anyways. As long as I get a healthy 4.4 - 4.5ghz overclock with decent temps, I will be happy.
I'm going to get the 3770k but not because of performance needs over my 2600k; rather, because I'm really curious to play around with an IC that has new 22nm 3D xtors under the hood.
Piledriver.
(Can't believe it took this long...)
I am actually interested in Trinity, if only for the PD preview.
I for one can't wait for Ivy to get here. I'm halfway through compressing a video in Sony Vegas Pro 10 that is taking 8 hours. I have some audio mixing to do but can't do it until the video is finished because I need the cpu cycles for the audio application to playback the mix. Presonus Studio One is what I'm using.
I know I should have gone for the 2600k instead of the 2500k but now it seems as though the smart thing to do is wait for the 3770k and it won't be here soon enough!
*snip*
atleast you get to test drive all of the newest cpus.Not a discount. This is my work computer and they would like me to have a new system with each generation so as to make sure I have experience with them when talking to you guys. So I already have ordered all the parts and am just waiting to get them and put it together.
At home I have an Intel® Core i7-860 that I would love to upgrade but right now the budget just wont allow me to do that.
What you say is great and all, but IB-E is supposed to bring eight-core CPUs into the consumer market. That, along with the IPC improvements, means a significant increase in multi-threaded performance in comparison to Haswell.
Ivy Bridge-E will definitely happen. If it didn't, there would be almost no reason to buy into Intel's high-margin Enthusiast platform.
if it's going to be way later than the IB launch they might as well just skip it and go with haswell-e