Haswell model specs leaked

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Well, if this slide is correct, I dont see it as a good thing. I would be happy to see the TDP go up if the chip was suddenly clocked to 4 ghz or something (or if it had more cores). But if the TDP goes up only to give better GPU performance, on the desktop I consider that a step back. I cant imagine buying a chip this powerful and not adding a discrete card. If the increased TDP is because of IPC increases or integrating more of the MB functions, it could be ok as well.

But overall, the more I see about Haswell, the less impressed I am for it as a desktop processor. Just depends on the IPC increase I guess, but I was hoping for an increase in base clocks and maybe turbo to 4+ ghz, or if keeping the same performance, even lower power consumption.
Certainly the measure of performance isn't based upon clockspeed alone. The cores themselves could have been improved which leads to improved performance. I think Intel is on the right track with having the GT2 on high end CPUs rather than GT3, gives more headroom for CPU performance. For desktops, I'd go with more performance with slightly higher power consumption. Added link to my news compilation, under rumors. :whiste:
 
Aug 11, 2008
10,451
642
126
Certainly the measure of performance isn't based upon clockspeed alone. The cores themselves could have been improved which leads to improved performance. I think Intel is on the right track with having the GT2 on high end CPUs rather than GT3, gives more headroom for CPU performance. For desktops, I'd go with more performance with slightly higher power consumption. Added link to my news compilation, under rumors. :whiste:

What I was trying to say was that I was hoping for increased IPC and an increase in clockspeed, or even more cores for the mainstream (I actually gave up on that long ago).
 

Borealis7

Platinum Member
Oct 19, 2006
2,901
205
106
so HD4600 is the name for GT2 graphics? and all i5's and i7's have that?

i would love a 4670K if it doesn't heat up like Ivys with additional voltage.
otherwise, it's a used 2500K for me.
 

Lepton87

Platinum Member
Jul 28, 2009
2,544
9
81
What I was trying to say was that I was hoping for increased IPC and an increase in clockspeed, or even more cores for the mainstream (I actually gave up on that long ago).
That's what I was hoping IVY would be, an increase in both IPC and clock-speed, overclocking headroom to be specific. Instead Intel cheapened out on solder and we got an increase in IPC alright but a regression in clock-speed headroom under non-extreme cooling. Even after bare die cooling it still does not reach the same frequencies as sandy.
 

moonbogg

Lifer
Jan 8, 2011
10,731
3,440
136
Enthusiasts will pay because that will be the only way to play.


If Intel starts charging $500 for a decent i7, you can consider me a former enthusiast at that point, at least where purchases are concerned. I'll go back to the good old days of getting more value out of a cheaper chip. I just won't spend $500 or more on a decent CPU. Won't do it. (as I ignore the 3930k in my case)

EDIT: I am not expecting anything exciting in the CPU area for a good long while. We got a real treat with sandy bridge, but I wouldn't expect that kind of perf jump to be anything like predictable in the foreseeable future. GPUs on the other hand just keep getting faster and faster at a pretty steady pace.
 
Last edited:

Fjodor2001

Diamond Member
Feb 6, 2010
4,061
465
126
So Haswell is expected to have 84 W TDP compared to 77 W for IB. I.e. 7 W difference.

Why do some people find that so hard to believe?

* Integrated VRM
* 4 more EUs
* Somewhat bigger CPU cores due to AVX2/TSX/etc

Wouldn't it be possible that this adds up to 7 W?
 

hokies83

Senior member
Oct 3, 2010
837
2
76
As long as it Avg % over Ivy clock for clock is 10% and it overclocks like a monster " it should: 4770k will be mine.
 

moonbogg

Lifer
Jan 8, 2011
10,731
3,440
136
As long as it Avg % over Ivy clock for clock is 10% and it overclocks like a monster " it should: 4770k will be mine.

It will be mine as well, so long as I can get 70fps instead of 60fps (that would be siiiiick).
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
Rumor up to now has been that desktop Haswell would go up to 95W, so this actually undershoots that. The rest isn't comparable because the mobile chips aren't listed.

7W at the same clocks is actually pretty modest given what Haswell brings to the table.. I bet it's a little more efficient overall than IB since Intel has had more time to optimize it for the process. Clock speed staying the same wouldn't be that surprising; the additional execution ports is only going to make scheduling need more cycle time so they may already be eating into timing headroom (meaning it might not overclock that well)

Oh please, no one is talking about these extremes.

10% higher IPC is okay, but CPU-wise one would expect a total of at least 20% higher performance as a product of IPC and clocks. We have seen that with Nehalem/Lynnfield and with Sandy Bridge. Compared to that, Ivy and Haswell are just disappointing.

Yeah, and before Nehalem we (generally) saw bigger still increases in performance between generations and moderate improvements in clock speed throughout the generation's life span. You didn't expect single threaded performance to keep improving at a steady rate forever, did you? There's a very real frequency wall Intel is staying behind; power consumption goes up a lot on the other side of it..
 

moonbogg

Lifer
Jan 8, 2011
10,731
3,440
136
You didn't expect single threaded performance to keep improving at a steady rate forever, did you? There's a very real frequency wall Intel is staying behind; power consumption goes up a lot on the other side of it..


That sounds pretty deep and epic. Could you elaborate on what this wall is about?
 

buklau

Member
May 4, 2012
135
0
76
I wonder how well it overclocks and will Intel use the el-cheapo thermal paste again this time instead of solder...
 

TuxDave

Lifer
Oct 8, 2002
10,571
3
71
That sounds pretty deep and epic. Could you elaborate on what this wall is about?

I don't think a "wall" is an appropriate term. It's closer to the idea that you can find better perf/power tradeoffs. If you can look at incremental gains. What does it take if you do nothing except bump 1% frequency. If I had to fling out a random number it could be something like 1% perf bump for 3% power bump if you do nothing besides just re-converge your product. So all features have to at LEAST beat that because that's the "do nothing" perf bump. And if you do find stuff, basically you start eating into your thermal envelope but getting more IPC than a simple frequency bump. So as long as you keep finding better stuff to fill in the thermal envelope, you lose simple frequency increases.
 
Last edited:

moonbogg

Lifer
Jan 8, 2011
10,731
3,440
136
I don't think a "wall" is an appropriate term. It's closer to the idea that you can find better perf/power tradeoffs. If you can look at incremental gains. What does it take if you do nothing except bump 1% frequency. If I had to fling out a random number it could be something like 1% perf bump for 3% power bump if you do nothing besides just re-converge your product. So all features have to at LEAST beat that because that's the "do nothing" perf bump. And if you do find stuff, basically you start eating into your thermal envelope but getting more IPC than a simple frequency bump. So as long as you keep finding better stuff to fill in the thermal envelope, you lose simple frequency increases.

So its a delicate balancing act that requires and increasingly incredible amount of creativity to slightly increase performance each generation. What i'd like to know is, what could they do if they had a much wider power envelope to work with? What i'm saying is like, what could they give us if they had a 500watt CPU socket envelope to work with for the enthusiast class CPU? I'd buy that, whatever it is.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
So its a delicate balancing act that requires and increasingly incredible amount of creativity to slightly increase performance each generation. What i'd like to know is, what could they do if they had a much wider power envelope to work with? What i'm saying is like, what could they give us if they had a 500watt CPU socket envelope to work with for the enthusiast class CPU? I'd buy that, whatever it is.

500W would give you something like 6-8Ghz I guess. Not exactly great to put it mildly.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
Is it really rational to complain about a 45 watt chip that will probably outperform a moderately overclocked 2700k? I'm talking about the 4770T of course. When playing most games, this thing is going to spend most of its time with turboing up over 3GHz. I'll reserve my disappointment for when a 4770T cant beat a stock cooled overclocked 2700k in games. If it cant do that then the uA is a fail.
 

Fjodor2001

Diamond Member
Feb 6, 2010
4,061
465
126
500W would give you something like 6-8Ghz I guess. Not exactly great to put it mildly.

All the way from the 1980--2005 we saw steady increases in CPU frequency for each new CPU generation. This was responsible for much of the performance increase during that period.

For example we went from:

Pentium II 233 Mhz (May 7, 1997)
Pentium II 450 Mhz (August 24, 1998)
Pentium III 800 MHz (December 20, 1999)
Pentium 4 2.0 Ghz (August 27, 2001)

So looking at the CPU frequency alone, the performance increased by almost 1000% in about 4 years! :awe:

Sure there may be bottlenecks and uArch differences so it's just an approximation, but still...

What is the reason the CPU frequency has topped out at current levels? Is there any physiological explanation?
 
Last edited:

inf64

Diamond Member
Mar 11, 2011
3,884
4,691
136
Is it really rational to complain about a 45 watt chip that will probably outperform a moderately overclocked 2700k? I'm talking about the 4770T of course. When playing most games, this thing is going to spend most of its time with turboing up over 3GHz. I'll reserve my disappointment for when a 4770T cant beat a stock cooled overclocked 2700k in games. If it cant do that then the uA is a fail.
Overclocked 2700K to what clock exactly ? If it is 4.5Ghz then stock 4770T has no chance outperforming it. OCed mildly to 3.7Ghz there is a great chance it will match/outperform it.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
So from SB Igpu to Ivy IGPU we have a 30% performance increase . In IVY bridge we have 20EUs (INTEL HASN"T SAID) in Haswell . From SB to Ivy bridge we have small ipc increase along with IGPU increase but a better power efficiency with Haswell that all goes out the window with Haswell. NOT likely even with more on die. Seems to me the people saying this leak is real and bad mouthing intel are the same people who were talking up BD Even tho reliable people told them otherwise. Were is the 125 watt unit intel said there would be . This doesn't look like a Intel graph at all. I haven't seen or heard anything about haswell , Robert would never tell me . LOL
 
Aug 11, 2008
10,451
642
126
Is it really rational to complain about a 45 watt chip that will probably outperform a moderately overclocked 2700k? I'm talking about the 4770T of course. When playing most games, this thing is going to spend most of its time with turboing up over 3GHz. I'll reserve my disappointment for when a 4770T cant beat a stock cooled overclocked 2700k in games. If it cant do that then the uA is a fail.

That seems like a pretty lofty expectation for a 45 watt chip. Not sure it can stay within the 45 watt envelope and turbo to 3+ghz, especially in a game that uses 4 cores.

So basically, maybe oversimplified, but Haswell is using about 10% more power for no clockspeed increase and maybe 10% better IPC. Where is the great increase in performance per watt that I thought Haswell was supposed to bring? Isnt this supposed to be a tock with major improvements in performance?
 

inf64

Diamond Member
Mar 11, 2011
3,884
4,691
136
Pretty sure moderately overclocked excludes 4.5GHz overclocks...

Turbo of 2700K is 3.9Ghz... That's only 15% less than "high" OC of 4.5Ghz. And SB core is very easy to OC even on air cooling. WHen you come to think of it, if one "OCs" the 2700K to 3.7Ghz with Turbo disabled there is a good chance his scores in many workloads will be lower/worse than what stock 2700K can do.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |