Just going to call this right now

Mar 10, 2006
11,715
2,012
126
Intel seems to be slacking and AMD seems to be picking up the pace. AMD's IGPs already smoke Intel's for GPGPU stuff and I don't see this lead shrinking. And given how lazy IVB seems and how Haswell's claim to fame will be its new instruction set (which AMD can implement), I think AMD has a solid chance of matching or even overtaking Intel.

Steamroller v.s. Haswell has a solid chance of being settled in favor of AMD. Thoughts?
 

Don Karnage

Platinum Member
Oct 11, 2011
2,865
0
0
Intel seems to be slacking and AMD seems to be picking up the pace. AMD's IGPs already smoke Intel's for GPGPU stuff and I don't see this lead shrinking. And given how lazy IVB seems and how Haswell's claim to fame will be its new instruction set (which AMD can implement), I think AMD has a solid chance of matching or even overtaking Intel.

Steamroller v.s. Haswell has a solid chance of being settled in favor of AMD. Thoughts?

If piledriver improves IPC by 15% i'll agree with you but as for right now you'd be a fool to buy Amd over Intel.
 

BD231

Lifer
Feb 26, 2001
10,568
138
106
monkey-stupid.gif
 
Mar 10, 2006
11,715
2,012
126
If piledriver improves IPC by 15% i'll agree with you but as for right now you'd be a fool to buy Amd over Intel.

Of course, right now buying AMD CPUs makes no sense if you're not using the iGPU, but I just think Intel's becoming stagnant, at least compared to AMD, who are trying their hardest to fight back. IVB on mainstream should've been a bigger improvement -- either more cores, more cache, etc.

Intel can keep the lead, but I'm worried that they're just going to let it slip.
 
Last edited:

Blades

Senior member
Oct 9, 1999
856
0
0
Right now, there is nothing comparable to Quick Sync over at AMD. If I were buying a laptop with an iGPU, I'd go with intel just because of that.. Not being able to play 1080p (60fps) with an AMD iGPU is a deal breaker. This is just keeping in mind future proofing the setup. Plus, when it comes to Linux - Intel are a bunch of sweethearts compared to AMD.. Granted, theres no quicksync support yet.. but still.. AMD's proprietary graphics driver is always a step behind and usually introduces a regression alongside an advance..
 

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
If they improve single-threaded performance by ~50%, yea. But I don't see that happening...ever. The IPC lead Intel has will continue due to fab advantage and AMD will need to pick up the pace elsewhere -- graphics and GPGPU is their main focus -- in order to close the gap.

I think the point you're making is interesting, though. The biggest jumps in performance from SB > IB > Haswell have been/will be attributed to graphics and not the CPU itself. That too will slow down and we'll be getting marginal increases in on-die graphics just as we do CPU performance and discrete GPU =P
 

Grooveriding

Diamond Member
Dec 25, 2008
9,145
1,323
126
Doubtful. How will AMD surmount Intel's superior in house process tech using a 3rd party foundry that is behind Intel's progression ?

It would be good for all of us no doubt. Would force Intel out of the ho-hum 15% a year and push them to offer more most likely, like we saw with Conroe. Still doubtful imo. Big time.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
If IB is so-so then AMD has a good opportunity to push farther into mobile and regular desktops. CPU power has been good enough for less savvy users since at least Phenom II/Core 2 and now it's about power+heat/cost/GPU.

Unless some high CPU demand app crops up for mainstream users.
 

Obsoleet

Platinum Member
Oct 2, 2007
2,181
1
0
If IB is so-so then AMD has a good opportunity to push farther into mobile and regular desktops. CPU power has been good enough for less savvy users since at least Phenom II/Core 2 and now it's about power+heat/cost/GPU.

Unless some high CPU demand app crops up for mainstream users.

I said this time and time again during BD's launch. Wonder why so many are now parroting it. It's the truth, but all I could find was hatred for Bulldozer when it was underwhelming, it was STILL more than 90% of users need.

In fact, 90% of users are happy with Apple A5 devices (sandboxed, not slow, works daily). Only content creators and the 2% of market that are nerds like us need this kind of horsepower.

My next PC, which I no longer buy any Windows devices besides my main desktop BTW, everything else is some sort of ARM based device (naturally crap Android phone, jailbroken ATV)- might very well be AMD if the iGPU closes in on stock 5870 performance in a few generations.

Could care less about insane i7 "performance", that's the last bastion of those who have dedicated their lives to Intel (and I love Intel products myself but I'm not about to be loyal to them).
 

Puppies04

Diamond Member
Apr 25, 2011
5,909
17
76
And given how lazy IVB seems

Yeah intel just designed and implemented a brand new transistor design, the likes of which have never been seen in a cpu while maintaining their frankly huge single threaded lead over AMD, all intel is going to do over the next few years is refine this technology.

Lazy fu**ers!!
 

frostedflakes

Diamond Member
Mar 1, 2005
7,925
1
81
IB isn't as bad as people are making it out to be IMO. Overclocking results so far are pretty underwhelming, enthusiasts who already own SB aren't going to be tripping over each other to sell off their SB chips and upgrade, but in pretty much every other way it's an improvement over SB. Smaller die that is easier and cheaper for Intel to manufacture, 5-10% IPC increase depending on the workload, lower power consumption which will allow them to better push their ultrabook form factor that they see as crucial to compete with thin, light, and long battery life ARM based devices like tablets, and better GPU performance (although it still can't quite keep up with Llano, much less Trinity, it is a big improvement over HD 3000). The only people who are really disappointed are OCers who were hoping these chips would easily hit 5GHz+ on air.
 

Tsaar

Guest
Apr 15, 2010
228
0
76
IB is a huge transition due to the physical limitations when dealing with transistors so small.

I am thinking of it as a beta test for Haswell based on the incredible SB architecture.

So yes, from an enthusiasts standpoint it doesn't seem that good, but hopefully we'll reap the benefits with Haswell.

I personally decided to go with SB due to incredible MC savings, but I am hoping IB serves its purpose.

Also...while I don't have specific numbers, the enthusiasts market has to be less than 5% of their revenue. For ultrabooks, IB is looking incredible.
 

CTA4LC4PON3

Member
Jul 21, 2009
140
0
0
I will NEVER buy amd. Intel ftw. Sorry but thats my opinion right there. I will never support amd as a company period. Intel will keep going UP & UP while AMD try's to play catch up.

the Rep over at overclock.net stated before bulldozer was released that AMD has no plans competing with Intel. even if amd does happen to pull ahead by a hair intel will pull ahead for the win
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
I will NEVER buy amd. Intel ftw. Sorry but thats my opinion right there. I will never support amd as a company period. Intel will keep going UP & UP while AMD try's to play catch up.

the Rep over at overclock.net stated before bulldozer was released that AMD has no plans competing with Intel. even if amd does happen to pull ahead by a hair intel will pull ahead for the win


Such attachments to a corporation aren't really healthy. They're pretty silly.
 

BD231

Lifer
Feb 26, 2001
10,568
138
106
As has been stated before node shrinks are getting more complicated, amd and nvidia go through the same issues.

Ivy shrink kinna sucked for the high end just like bulldozer, just like fermi. Nvidia worked out the kinks, amd fired everyone, and intel will keep commiting to improving upon weakpoints while maintaining total dominance. I dont see how going through the paces is stagnating.
 

gevorg

Diamond Member
Nov 3, 2004
5,070
1
0
AMD has a long way to go to match Intel, the FX-8150 consumes power roughly as much as Intel extreme i7-3960X, while performs behind midrange 2500K/2600K in most benchmarks.

Anandtech
42358.png


Xbitlabs
power-2.png
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
You realize that the 8150 is pretty comparable to an i7-920? That launched in late 08, fully 3 years before the 8150.

You think they will make up a 3 year gap in a year and a half? That type of magical thinking is what got people excited about BD in the first place.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
The entire industry is stagnating when it comes to performance. Intel for Haswell is saying the best of use of their transistor budget is to bring the NB elements onto the CPU core. That does kind of suggest they can't utilise them to improve the performance of the CPUs calculation or its no longer the priority.

Its not just Intel, everyone is struggling with the physical limitations they are running into. I don't think Intel's problems are indicative that AMD can catch up, I think if anything it suggests that AMD might fall further behind as the expertise to implement the technology rises.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
1. Power efficient design is a priority

Designing for performance without caring about anything else can't happen anymore. CPUs already run at 130W or more. In the days of rapid performance increases(for both CPU and GPU), the power usage were really low.

Only a little over a decade ago, Geforce 2's consumed less than 30W and Pentium 4's going over 50W was a big surprise. Moore's Law allows power reductions, but designs have gone way above that each and every generation. Claims made by companies like Nvidia saying their GPUs are advancing at 3x the speed of Moore's Law wasn't being done for free. It wasn't going to last.

Nowadays we say we care about "power reductions" but in the Pentium MMX days the HSF combos were barely larger than what we use in chipsets. Few years before that, active cooling weren't even needed.

2. Workload benefits are plateauing, or becoming really difficult to optimize

Back in 2005 when first dual cores were surfacing, the benefits were throughout user and application base. We dreamed of multi-core era lasting forever and single thread IPC gains seemed no longer important. That's not the case anymore. When we moved to quad cores, it took significantly longer to take advantage of its gains. With 6 cores, its even less than that.

3. Physical limitations?

In the Athlon and Pentium III days, the war between two companies AMD and Intel were basically all about clock speed. Then few years later, Intel predicted 10-20GHz computers in 2015+.

Nowadays the record for highest speed is only set in overclocking. But not even with liquid nitrogen cooling, voltages unimaginable to mere mortals, with cherry-picked parts can we reach 10GHz. It makes me wonder if its even possible to see 5GHz chips at stock at all.

That does kind of suggest they can't utilise them to improve the performance of the CPUs calculation or its no longer the priority.
Remember what Anand said on his reviews? It used to be CPU architects did quite a bit to get that extra 1% in performance. But with newer architectures, the circuits are only used if 2% performance gains happen with only 1% increase in power use.
 
Last edited:

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
GloFlo is no where to be seen with their FinFET implementation and yet Intel is bringing out theirs in a few weeks time, I doubt Intel is anywhere near being lazy. From looking at how much the IGP has progressed from SB to IB and down to Haswell, Intel is actually taking IGP seriously this time and I can see that benefiting a bigger majority of users, mobile and desktop.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
Llano pretty much made up all of the low power ground Intel had on AMD.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
I say by 2020 they will start with a 1nm process ,,,,,,,,, The CPU is soo fast, it does the work for you,, you just sit there and command it and it does it lol

;)
 

Mxster

Member
Mar 25, 2012
87
0
0
I say by 2020 they will start with a 1nm process ,,,,,,,,, The CPU is soo fast, it does the work for you,, you just sit there and command it and it does it lol

;)

I'm just replying 2 the guy above me, wtf does this even mean. Lol
 

nismotigerwvu

Golden Member
May 13, 2004
1,568
33
91
I say by 2020 they will start with a 1nm process ,,,,,,,,, The CPU is soo fast, it does the work for you,, you just sit there and command it and it does it lol

;)

I can assure you that you're not going to see a silicon process down that small (at least the way we use them now). I don't know the atomic radius of Si off hand, but 1nm would be somewhere around 9~10 atoms. Also, you'd likely have to use AFM to etch features that small, I just don't see lithography reaching 1nm with useable yields for a long time.