Average Performance Delta between Phenom II, Core 2, Core i7, and Sandy?

[BoomerD]

As pointed by your post s text and your expressions, porn
sites suit you way better than IT forums....
But for your insight, go in the old pages and check s
anand s P4 review that ended being rewrited a few day
later with Intel sending a full crew to recompile the softs
before publication...

I'm curious...if you think Anandtech is so biased and under the control of Intel, why are you, as an AMD "enthusiast" hanging around here?
Obviously, everyone is welcome to have their own viewpoint and opinion about the hardware, but you're not only slamming Intel, but the site here as well.
If you don't like it here, you're certainly free to log out and not return.

As I said earlier, I don't think AMD processors are BAD, but they're just not GOOD.
Maybe I should qualify that a bit since you have a problem understanding my point...

At the price point that AMD is selling their processors, they're not competitive with the Intel offerings. They DO offer some good value at a much lower price point, for use in budget machines that aren't aimed at hardcore computer gamers, but dollar for dollar, they just don't come close to matching the performance of the Intel processors.

FWIW, I'm actually glad that AMD is around and still participates in the processor wars. Without them as competition, Intels processors would probably get considerably more expensive.

 

[Abwx]

I'm curious...if you think Anandtech is so biased and under the control of Intel, why are you, as an AMD "enthusiast" hanging around here?
Obviously, everyone is welcome to have their own viewpoint and opinion about the hardware, but you're not only slamming Intel, but the site here as well.
If you don't like it here, you're certainly free to log out and not return.

As I said earlier, I don't think AMD processors are BAD, but they're just not GOOD.
Maybe I should qualify that a bit since you have a problem understanding my point...

At the price point that AMD is selling their processors, they're not competitive with the Intel offerings. They DO offer some good value at a much lower price point, for use in budget machines that aren't aimed at hardcore computer gamers, but dollar for dollar, they just don't come close to matching the performance of the Intel processors.

FWIW, I'm actually glad that AMD is around and still participates in the processor wars. Without them as competition, Intels processors would probably get considerably more expensive.

Too much things at the same time, but i ll try to answer without
missing the essentials..
Contrary to your sayings, i m not an AMD fan , but let s say that
Intel has, for some reason , adopted a very unethical policy for
many years and that policy was aimed at screwing the consumer
and force him to pay them margin not seen in any other technical
business.
As you pointed it, without AMD, we would pay our processors
something like 1000 to 2000$...
I praised the Pentium Pro perfs, but not its price, and i invite you
to check what was the CPU prices from the pentium1 to about 1999,
when the first athlon was launched (at 550$).

As for price perf ratio, you sound contradictory.
According to your sayings, an intel CPU has better perf/$.
So if we take a X6 at 200$, a 300$ intel CPU should have
more than 50% better perfs to deserve the perf/$ crown,
wich is evidently not the case.
We could take a 100$ A2X4 exemple and it would lead to
the same conclusion.
At each price point AMD has better perf/$ , anyone here
will tell you the same.
Do the math, only add/subs/mult/div are necessary...

Indeed, your POW just prove how an efficient marketing can
inflate the real numbers .

 

[Chaoticlusts]

As entertaining as reading the arguments with Abwx is, would there be any chance of this thread getting back on track? the OP's question was quite interesting, finding proper performance comparisons for old products seems a little hard sometimes >_< (I'm running an original E6600, I honestly have no clue what that equates too in terms of stuff on the market atm..I'm assuming mebbe a $40-50 cpu ) can be very useful if your planning to sell your old system (assuming you want to sell it for a fair price..if you want to screw people just make up whatever sounds good )

 

[IntelUser2000]

I think when taking OP's question into context, the 5-10%(or even 20%) differences in IPC is not so important. He wants to know them for laptop comparisons. It's true while Core 2 depending on the variant has advantage over Phenom II, the clock speed in desktops are higher for the Phenom II as well.

Phenom II's are so behind on mobile that the quad cores are playing in the Core 2 Duo territory: http://www.anandtech.com/bench/Mobile/148

Need a new laptop after old one died from being dropped in a puddle. Borrowed a Pentium M from the office but it's slow. I was thinking about getting a quad core Amd for 500$(refurb).
Or I can drop another grand for a brand new sandy bridge laptop. Either way I need something for light gaming + number crunching.

(Or he's really doing that to make excuses that he is in the wrong thread but let's assume he's not)

 

[dac7nco]

Chaoticlusts, For your E6600 you could cut the transcoding numbers of the Q8300 roughly in half.

Abwx, the point is that AMD has no CPU worth spending >$100 on, not that they don't compete below that. An i5-2500K is much superior CPU compared to AMD's X6 1100BE, is a bit cheaper, and will effortlessly break 4.4GHz with a $30 cooler on the cheapest P67 board you can find. An i7-2600K is not only the fastest retail quad-core, but the fastest tri-core, dual-core and even single-core CPU in existence.

My personal experience is in very large data centers, both with highly multi-threaded tasks and with "narrow" tasks which scale only with clock speed. With the rare exception of the highly parallel Magny-Cours (G34) Opterons under very few workloads, Intel has been a better solution since Nehalem (2.5 years). On the desktop, SB>Westmere>Nehalem>Core-2=K10.5.

Daimon

Edit: I believe you may be considering the AMD 6-cores and Sandy Bridge as "high-end" CPUs. AMD doesn't have any of those anymore, and Sandy Bridge started with the mid-range first; Sandy Bridge-E will be Intel's high-end late this year. The soon to be released FX is going to be priced to compete with Intel's new mid-range.

 

[RobertPters77]

Well abwx you succeed in driving me away from an amd purchase.

Fuck the underdog. I'll go support the 'Evil Empire.'

 

[Abwx]

Well abwx you succeed in driving me away from an amd purchase.

Fuck the underdog. I'll go support the 'Evil Empire.'

Keep cool Robert, acting under emotion is always counterproductive...
To aggravate my case, i ll point that i m posting using my old and
beloved first class (at the time) laptop wich has...an Intel cpu !!,
and worse, it s a ( Northwood) P4M...
I did buy it at the time after reading Anand s
praise of this model. ( a Compaq)..

 

[Abwx]

Abwx, the point is that AMD has no CPU worth spending >$100 on, not that they don't compete below that. An i5-2500K is much superior CPU compared to AMD's X6 1100BE, is a bit cheaper, and will effortlessly break 4.4GHz with a $30 cooler on the cheapest P67 board you can find.
Edit: I believe you may be considering the AMD 6-cores and Sandy Bridge as "high-end" CPUs. AMD doesn't have any of those anymore, and Sandy Bridge started with the mid-range first; Sandy Bridge-E will be Intel's high-end late this year. The soon to be released FX is going to be priced to compete with Intel's new mid-range.

I wouldn t go as far as saying that they have nothing over 100$
that is worth buying.
Seems to me that at 100$ , the A2X4 is a convenient CPU, and
that the X6 suit well when multitasking is the need.
True that between those two, there s nohting substancial,
neither do they have any product that can compete with
Intel s fastest offerings.

As for the BD, it s still an unknown, but i doubt it will
be priced as high as a 980/990X even if it is faster.

As for SB, it will be upgraded to 6 cores to be competitive with BD.
We have no benches yet, but Intel s move is a good indication
of what to expect, since they already know how the competitor s
product will perform..

 

[Morg.]

IPC will always matter. Core count is a strong second but IPC will always matter. Answer me this. If core count is so important then why isn't AMD competitive with intel? 100$ Quad core, 200$ hexacore. They should be making a killing.

They actually are making a killing ... check the server market some time, and also server benchmarks. good progress for AMD it was.

 

[Morg.]

My personal experience is in very large data centers, both with highly multi-threaded tasks and with "narrow" tasks which scale only with clock speed. With the rare exception of the highly parallel Magny-Cours (G34) Opterons under very few workloads, Intel has been a better solution since Nehalem

I respectfully disagree.. Until very recently there was NO 4-socket system from Intel that could beat a quad socket opteron. - Luckily that's changed.

 

[dac7nco]

Intel, until Magny-Cours, had 2P systems which matched the 4P Opteron. A Pair of 5570s in 2009 was also a lot cheaper than four Shanghais. My Tyan Thunder 4P workstation lasted me all of five months until Tylersburg/5520 based machines ass f*!@ed it.

Daimon

Edit: Morg., if you're talking about Intel's Nehalem-EX, I wouldn't put it into the same class as Magny Cours; If AMD has done one thing I'm fond of, it's removing the "4P tax" from G34. a S1567 Intel will cost you your first born.

 

[tweakboy]

You might get better response from the SFF and notebooks forum, but I will leave this here, as it is CPU questions.

Ill tell you fact as said in AIDA a AMD 6 core It beats it by a large number. I due some Queen and FPU tests,, 6coreamd loses even tho it has 2 more cores....

So we have learned its not all about the cores,, rather the core speeds.. thx gg and gl

 

[intangir]

They actually are making a killing ... check the server market some time, and also server benchmarks. good progress for AMD it was.

Er, no. No, they aren't. As of the middle of 2010, AMD's server marketshare was much weaker than its share in the desktop or mobile market.

http://hothardware.com/News/IDC-Claims-AMDs-Server-Market-Share-Fell-33-Percent-Last-Quarter/

If AMD's server sales were indeed within the company's expectations, it can only mean the company was expecting to be pummeled. According to IDC's figures, Intel finished Q2 with 93.5 percent of the server market, up 3.3 percent, while AMD's share decreased to 6.5 percent, down by the same amount. That's rather depressing news considering AMD's 12-core Magny-Cours (and dramatically reduced prices) were available through the entirety of Q2 with Lisbon popping up right at the end of the quarter.

Server: 6.5%
Desktop: 27.3%
Mobile: 13.7%

I'd love to see more recent numbers, but the trend is clear. This is a far cry from the 33% server market share AMD peaked with back in 2006, in the days of the Opteron before Woodcrest! And looking at the benchmarks, it's easy to see why its fortunes have taken a turn for the worse. Intel's current Westmere Xeons blow away AMD's offerings in both performance and power efficiency.

http://techreport.com/articles.x/19196/8
http://techreport.com/articles.x/19196/6

If that's making a killing, I'd hate to see what you would consider "not doing so well".

 

[OCGuy]

Wow I was not aware of just how bad AMD is doing in the server market.

 

[Topweasel]

IPC will always matter. Core count is a strong second but IPC will always matter. Answer me this. If core count is so important then why isn't AMD competitive with intel? 100$ Quad core, 200$ hexacore. They should be making a killing.

IPC is bull crap. Its a catch phrase used to explain to people to stupid to research their purchases on why their new machine with less GHz is faster then their old machine. There are dozens of ways to make a better processor and IPCPC is a very legitimate approach but its not the only approach. For example if someone can keep the IPCPC within 15-20% of Intel and get 10% faster CPU and double up on cores, then that can be as much a winner.

AMD's Phenom II isn't make a real killing for several reasons, architecture reasons (not limited to IPC), debt reasons, anti-competitive reasons, production reasons, name reasons.

 

[dac7nco]

Some of these people are akin to religious zealots; OP started this thread wondering about performance improvements in CPU generations, and it was immediately derailed into "Kia makes a better car than BMW".

Failure.

 

[Topweasel]

Some of these people are akin to religious zealots; OP started this thread wondering about performance improvements in CPU generations, and it was immediately derailed into "Kia makes a better car than BMW".

Failure.

No it was a troll thread don't want to point figures but I am sure looking at some of the posters particularly some that started this thread you would notice that a decent amount of posts are users displaying accurate knowledge of the information they pretended to have no knowledge of in the first place and going back through their limited posting history shows a behavior of intelligent but subversive posting.

 

[RobertPters77]

@abwx

You were trying to steer me into purchasing an Amd chip. You failed. You're no better than the countless 'macheads' at my former place of work, who scream, bitch, and yell about how awful windows is, yet they can't show me one shred of proof that mac is superior. All they talk about is how evil microsoft was because it was keeping apple down in the 80's. Suddenly Company 'A' is the little guy and we should all feel bad that they are in that position? Second to EVERYONE. ************ There's a reason they're the 'little guy.'

*********** Boo-Hoo. Company 'X' prevented company 'Y' from selling it's product to retailer 'A.' Oh wooh are they. Have you ever heard of exclusivity deals and group discounts? Like you said. If intel gave that french retailer free advertising for signing an exclusivity deal? What's the problem? Likewise if Nvidia pays game developers to optimize for their line of geforce products? Then what's the problem? Is Amd somehow morally superior because it refuses to 'stoop down' to this level of competition which you define as 'unfair?' No they just are weak when it comes to making business decisions.

Intel as of now makes a better chip. And I'll buy it because it's better. If Amd releases a better chip then I shall buy Amd. I am not brand blind. I want the best performance bar-none. I currently have 4 Amd systems in my house + Another 2 Amd laptops. I bought them because they were what I wanted/needed at the time. I use for Business intents and purposes an Intel laptop because A. Clock for Clock it's faster. Which means, Time is Money. & B. Battery life. My old 720QM outlasted my Amd laptops. Even though the 720 system had much more power draining hardware.

@Topweasel
Wrong. Time is Money. A chip with strong IPC gets work done quicker = Time is Money. In my former place of work we had Amd Magnycours chips in limited server deployments for non essential tasks. Every other computer used Intel chips. Mobile, Desktop, and Essential Server.

@intangir
Wow that bad? Seems about right. I remember we had 1 Amd Magnycours server for every 20 Intel servers(And we had less than 200 servers).

 

[Abwx]

@abwx

You were trying to steer me into purchasing an Amd chip.

Everyone is entitled to have its own opinion, but on this one,
you are just plainfully wrong, since i just don t want to fall in
the same extremities as the firm i m condemning.

This is exactly what intel has done during the P4 era, particularly
at the beginning, explicitly , forcing the consumer to buy an
inferior product, and now, you are just boasting that you want
the better chip.
In my region, in 2001/2002, it was simply not possible to buy
a PC with the best chip, thanks to those infamous contracts.

 

[RobertPters77]

Basically you have a chip on your shoulder because Intel signed some exclusivity deals with retailers in your area?

So should I be mad at Macy's because the nearby mall signed a deal with them for 10 years exclusivity, in exchange for free advertising?

Or should I be mad at apple because I can't buy a macintosh with an Amd cpu? Nope.

P.S. Learn to spell.

 

[aigomorla]

The benches don't lie but I need more info.

overclocks do, so its very not recomended for you to ask.

Basically i can ramp up any of those cpu's to unsafe voltages, and OC them to butfungulio and back on my LC'd systems.

And show u the lowest out of the lot slaughtering the highest out of the lot.
(you think im kidding? ... well a C2D beating my 990X might be difficult since stock on the 990X is already up there in E6600 wall territory.)

So take members posts with a very big grain of salt.

If your looking for stock benches, well, go on AT's main forum, look up any CPU review, and they should all be listed.

 

[Abwx]

Basically you have a chip on your shoulder because Intel signed some exclusivity deals with retailers in your area?

So should I be mad at Macy's because the nearby mall signed a deal with them for 10 years exclusivity, in exchange for free advertising?

Or should I be mad at apple because I can't buy a macintosh with an Amd cpu? Nope.

P.S. Learn to spell.

What is condemnable is the fact that a firm with an inferior
product tried to bribe the whole distribution chain at the expense
of the consumer, and indeed, they succeeded.
Remeber the K7 launch with almost no MB manufacturer
wanting to support this plateform?..
Only crappy FIC and a little better MSI were produced
the six first months, before the i820 chipset epic failure
that almost sent them under chapter11 abruptly changed
the deal....

As for mispellings, well, english is not my mother tongue,
you should have guessed it (or not??)....

 

[Arkaign]

What is condemnable is the fact that a firm with an inferior
product tried to bribe the whole distribution chain at the expense
of the consumer, and indeed, they succeeded.
Remeber the K7 launch with almost no MB manufacturer
wanting to support this plateform?..
Only crappy FIC and a little better MSI were produced
the six first months, before the i820 chipset epic failure
that almost sent them under chapter11 abruptly changed
the deal....

As for mispellings, well, english is not my mother tongue,
you should have guessed it (or not??)....

A friendly tip, please don't hit the enter button after every few words, the forums have word-wrap enabled, so doing that just makes your posts look muddled and out of place.

I do remember the K7 launch, to be fair the first K7 chipsets were pretty terrible. It wasn't until KT266A and not long after Nforce2 that the AMD-compatible chipsets really started to pick up and fly. KT133A was fairly good as well, I remember it actually outperforming the early AMD 760 DDR chipsets using much cheaper Sdram memory in certain matchups, and never really being that far behind in real-world benches of any type. Kingmax tBGA PC150 hahaha.

It's not terribly surprising that the early K7 support was thin on the chipset side. Consider it from a manufacturer's perspective :

All previous AMD products were socket-compatible with Intel processors, meaning that there was little to lose by having the correct microcode and such to support the AMD chips as they came out. After all, ANY credible mobo maker's Socket 7 boards always supported the latest AMD chips at the time, and many added support via BIOS updates for chips as they came out. I remember being able to run a K6-2 300 in a very old Abit mobo, TX chipset IIRC. This was a very different scenario when Athlon Slot-A was launched, as there was considerable expense in developing the mobo, and if it turned out to be a failure like K5 for example, there was a ton of money to lose without the benefit of just being able to sell it also to Intel customers. In essence, you are taking out the vast majority of the existing market for customers just by making a Slot-A mainboard. You could also say it's an opportunity to serve more of the potential marketplace, which is correct, but it wasn't known at the time whether or not it would pay off. 3dnow was already a bust, even though it was pretty good tech in my opinion, and the K6 models had lost serious ground against cheap, easy-to-overclock Celerons and the like.

So I think the K7 mainboard availability at launch had a lot less to do with manipulations by Intel, and a lot more to do with the realities of the market at the time. Building a mainboard around 440BX gave you a huge number of potential buyers, including countless people that already had older 440FX and 440LX-based boards, and a proven set of processors that were selling extremely well. Building a mainboard around AMD760 was something of a gamble, zero existing customers, unknown qualities compared to the competition, and just a lot more risk. Luckily it paid off, the chip was great, and it led to VASTLY better chipsets down the line.

 

[Arkaign]

Too much things at the same time, but i ll try to answer without
missing the essentials..
Contrary to your sayings, i m not an AMD fan , but let s say that
Intel has, for some reason , adopted a very unethical policy for
many years and that policy was aimed at screwing the consumer
and force him to pay them margin not seen in any other technical
business.
As you pointed it, without AMD, we would pay our processors
something like 1000 to 2000$...
I praised the Pentium Pro perfs, but not its price, and i invite you
to check what was the CPU prices from the pentium1 to about 1999,
when the first athlon was launched (at 550$).

As for price perf ratio, you sound contradictory.
According to your sayings, an intel CPU has better perf/$.
So if we take a X6 at 200$, a 300$ intel CPU should have
more than 50% better perfs to deserve the perf/$ crown,
wich is evidently not the case.
We could take a 100$ A2X4 exemple and it would lead to
the same conclusion.
At each price point AMD has better perf/$ , anyone here
will tell you the same.
Do the math, only add/subs/mult/div are necessary...

Indeed, your POW just prove how an efficient marketing can
inflate the real numbers .

This is a hugely blind post. Whenever Intel or AMD has a huge performance lead, they bleed the heck out of the consumer. In fact, I'd say AMD is WORSE than Intel in this regard. I have a perfect example for you :

When AMD held the performance crown with Socket 939, and dual-cores were new, the CHEAPEST X2 model, the 3800+, was solidly around the $300 price point, with the top-end models all hovering from $500-$1000+. In essense, they pretty much put the screws to the public unless you wanted single-core, which were STILL fairly pricey. This held right up to Conroe's release, even though cheap (and obviously inferior) Pentium D models could be had for peanuts in comparison, and once overclocked held their own nicely with the bottom-end X2s.

When Intel held the performance crown, you at least see value chips (E6300, etc) that aren't complete wallet-destroyers, and were contemporary with the high end of the time. Quads were very expensive on 65nm no doubt, but even Q6600 became fairly affordable pretty fast, and cheap dual-cores became widespread with stuff like E4xxx series.

If you look at the case now, even a 2500K dominates everything AMD currently has outside of a handful of oddball benchmarks, and once overclocked, dominates it in absolutely everything. And how much does this total domination cost? A couple hundred bucks or so?

I will say that I think Intel priced S775 quads a bit too high for too long though. But the point remains, AMD wasn't nice at all on price when they had the crown, in fact they were terrible.

 

[itsmydamnation]

to make that a fair comparision, you would have to compare it to something like the low end I7 920 that would actually be comparable to the 3800 X2 in product line up. dont know what it was like where you are but in OZ they where about the same price (infact i think 920 was a little bit more, i would have to dig out my invoices).

I would also compare the 3800 X2 to base model of the SB's about to be released, and i would expect them to be at about the same price point again at launch.