waiting hashwell...

It will arrive in Q1 2013 if all goes right. That is not that soon to be honest.

22nm, 95watt TDP. Going by that we can assume it will be quite a bit faster than Ivy Bridge.
 
Also, do not think that there is a direct upgrade path from Sandy/Ivy Bridge to Haswell, there is a socket change over. Going from LGA1155 to LGA1150.
 
I believe that in general dry hash is preferred so storing it in a well is not ideal.
 
Wyluliraven said:
Also, do not think that there is a direct upgrade path from Sandy/Ivy Bridge to Haswell, there is a socket change over. Going from LGA1155 to LGA1150.
Click to expand...

Maybe ASRock can make another oddity of using a newer chipset with an older socket, much like they did with their P67 Transformer.
 
OP if you want to know about Haswell go on Google and type in: Intel Haswell news. All of the links that pop up, those links have all of the info that is currently known about Haswell.

Go do the research, then if you have any specific questions or comments make them here (please include a link so we can all read along :)). It's a waste of time for us to paraphrase everything again.
 
Well, I'll do a bit of paraphrasing :p

Significantly stronger iGPU.
Integrated VRMs. <--- VERY IMPORTANT CHANGE FOR US ENTHUSIASTS
Top model will be quad-core with hyperthreading.
Stronger CPU.
Rumored to have L4 cache for sharing between CPU and iGPU.
16 PCI-E 3.0 lanes
Dual channel DDR3
uses LGA1150.

I think that sums it up for what we know so far.
 
Yeah, Tsumi's right, it can't hurt too much to paraphrase: ;)

  • Rumor is that the fastest IGP will only be available on laptops (heard this on SemiAccurate). Apparently Intel and Apple want to secure the position of laptop graphics in comparison to desktop (AFAIK).
  • The integrated VRM might allow Intel to lock voltage modifications on non-K series processors. I've never heard anything about it but it could happen.
  • What's good about Intel right now is that the integrated PCIe lanes now have enough bandwidth to support tripleGPU with all but the fastest graphics cards. You can split the PCIe 3.0 ports into 8/4/4 and run three cards. With most of the Z77 boards there was alot of stratification with this one feature, only the highest boards in the lineup could do this. Hopefully with Haswell more budget boards will have three PCIe slots wired to the CPU. :cool:
 
Tsumi said:
Well, I'll do a bit of paraphrasing :p

Significantly stronger iGPU.
Integrated VRMs. <--- VERY IMPORTANT CHANGE FOR US ENTHUSIASTS
Top model will be quad-core with hyperthreading.
Stronger CPU.
Rumored to have L4 cache for sharing between CPU and iGPU.
16 PCI-E 3.0 lanes
Dual channel DDR3
uses LGA1150.

I think that sums it up for what we know so far.
Click to expand...

LGA1150? Where did you read it?
Is it a rumor or a confirmed info?
 
LGA1150 is confirmed. Although AM3+ is also a dead end also after this year and I believe that is a rumor.
 
sblantipodi said:
LGA1150? Where did you read it?
Is it a rumor or a confirmed info?
Click to expand...

Confirmed. Either way, it will be a different socket due to the integrated VRMs. There is absolutely no way Haswell can work with SB/IB boards.
 
Why do you people wish your life away in hopes of a product? LMAO...

I wish my life would go slower.
 
Rumor...

Haswell to be 10+ percent faster than Ivy Bridge

Intel internally calls Haswell its “fourth generation” Core processor micro architecture, and we believe that Intel counts only major steps in its development.

Its first Core architecture was called Conroe /Merom in 65nm, followed by Nehalem in 45nm, Sandy Bridge in 32nm and Haswell will be the second 22nm core with a new architecture.

Intel still plays its old tick-tock game and Haswell is a major step, or a tock, while Broadwell is 14nm shrunk version of the Haswell architecture. A more obvious example is Sandy Bridge in 32nm, a tock in Intel’s development cycle, while the most recent Core processor is a tick, and it is based on 22nm Ivy Bridge core.

Now Intel tells its partners to expect that Haswell should end up at least 10 percent faster than Ivy Bridge based cores at the same clock. These numbers are based on pre-silicon projections that Intel always does before it gets the working prototype back, but since we are some month away from the IDF 2012 in San Francisco, we are quite sure that we will see Haswell again, much closer to its final design, and we even expect to see Broadwell prototypes to hit at least one of the keynotes.

Intel also hints at enhanced based overclocking and end user tunability, which sounds promising to many enthusiasts. If all goes well for Intel, Haswell should be quite a nice piece of hardware once it ships in Q2 2013.
Click to expand...

Source: http://fudzilla.com/home/item/28318-haswell-to-be-10%20-percent-faster-than-ivy-bridge
 
Applications that can take advantage of AVX2 should be a whole lot more than 10% faster.
 
Transactional memory support (Intel is calling it TSX) will be huge boon to software developers. I am very excited about that feature.

Doubling the cache line size could have very tangible benefits as well.
 
Haswell smaswell.. im thinking of selling my signature rig, except my case labs case, and just using my new Mythlogic GTX680M powered 3720QM laptop that will arrive in two or three weeks muhaha.

My Laptop is going to be faster than a 2600K

At any rate I video transcode so much I am really hoping that Haswell IPC will be significantly faster than my Sandy Bridge E and if so I will adopt it over my Sandy E which is rather rediculous if you ask me in power.
 
Not with a base clock speed of 2.6GHz it isn't. and I'll bet the 3.7GHz turbo has a much tighter power envelope than a 2600k.Meaning it won't maintain that turbo ratio for anywhere near as long as a 2600k would.
 
10% higher IPC. That's a bit disappointing, and that relatively small increase should give AMD some time to catch up. Looks like Intel really is focusing heavily on the iGPU side.

I'll reserve judgement until we see how well it clocks. If clocks are similar to current offerings, then it's very disappointing. If clocks are at least 10% higher in the same thermal envelop, resulting in a 20% increase at the same wattage, then it's a decent step forward. If clocks are 20% higher...
 
and that relatively small increase should give AMD some time to catch up.
Click to expand...

I do not think that gives AMD much of anything. I mean if AMD has a 15% IPC improvement that would be a net gain of around 5% however they are behind in IPC by so far its not like 5% would eat at all into the lead.
 
Last edited:
DDR4 memory support for the Xeons?

Looks like Broadwell might follow with a new socket too.
 
I'm still rocking an 1156 setup. That sucks that there is going to be yet another socket change... I guess this is why I'm still using an 1156 setup and have not upgraded.
 
Atl530i said:
I'm still rocking an 1156 setup. That sucks that there is going to be yet another socket change... I guess this is why I'm still using an 1156 setup and have not upgraded.
Click to expand...

When you keep adding features to the CPU, it's impossible to keep the same socket.
 
Tsumi said:
10% higher IPC. That's a bit disappointing, and that relatively small increase should give AMD some time to catch up. Looks like Intel really is focusing heavily on the iGPU side.

I'll reserve judgement until we see how well it clocks. If clocks are similar to current offerings, then it's very disappointing. If clocks are at least 10% higher in the same thermal envelop, resulting in a 20% increase at the same wattage, then it's a decent step forward. If clocks are 20% higher...
Click to expand...

Intel has increased IPC with every new generation for the last many years. AMD has decreased it, with BD in particular. I'd like to know how this math works where AMD catches up...
 
drescherjm said:
I do not think that gives AMD much of anything. I mean if AMD has a 15% IPC improvement that would be a net gain of around 5% however they are behind in IPC by so far its not like 5% would eat at all into the lead.
Click to expand...

RamonGTP said:
Intel has increased IPC with every new generation for the last many years. AMD has decreased it, with BD in particular. I'd like to know how this math works where AMD catches up...
Click to expand...

Except AMD is also upping up their clocks quite a bit.

IPC isn't everything, how many times does that need to be said? AMD aimed for a high clocking design. As a result, they sacrificed some IPC for high clocks, but they failed to achieve their high clock target.

Actually, that technically is wrong. Intel has been increasing IPC every generation because each generation is based on the previous generation, but improved. AMD's drop in IPC was due to introducing a radically new and different design that was not based on anything it was producing before. Each subsequent generation will yield IPC gains, because now the focus is on improvement, not something new.
 
Tsumi said:
Except AMD is also upping up their clocks quite a bit.

IPC isn't everything, how many times does that need to be said? AMD aimed for a high clocking design. As a result, they sacrificed some IPC for high clocks, but they failed to achieve their high clock target.
Click to expand...

So they failed in clock speed and IPC, which means down on both multiples... Now you've got me wondering even more how this math is going to work out? Sorry, but AMD is not going to make any ground on Haswell.
 
RamonGTP said:
So they failed in clock speed and IPC, which means down on both multiples... Now you've got me wondering even more how this math is going to work out? Sorry, but AMD is not going to make any ground on Haswell.
Click to expand...

Maybe because AMD plans to improve their cores each year, while Intel only does so once every 2-3 years?

Seriously, it's almost as if you don't want AMD to try to improve and catch up. And thanks for totally ignoring the last paragraph, love how you guys like to pick out certain things.

And did you even bother reading? If Haswell is only a 10% IPC increase, it should allow AMD some time to catch up, since Vishera is a 10% IPC (approximate guess based on Trinity) + 10% clock speed (based on leaked 8350 specs) increase. And Steamroller, which the big changes are supposed to be happening, is set to be due out soon after Haswell.

If Haswell has both IPC and clock speed gains, that's a different story.
 
You're confusing my realistic outlook with not wanting them to improve. I'd love for them to improve. I just don't see how you're concluding that this is somehow a silver lining for AMD considering Intel has improved each generation (haswell is no exception if we are to believe preliminary reports) while AMD has either been stagnant or went backwards. I hope I'm wrong, I just don't think I will be. I see the gap widening, not narrowing unfortunately. I think the one advantage AMD has now is their iGPU and even that will be a thing of the past with Haswell while making up zero ground on the CPU front.
 
You must log in or register to reply here.
Back
Top