Is Intel really charging $300 for an extra 0.1 GHz?

rutski89

Honorable
Aug 11, 2013
28
0
10,530
Have a peek at Intel's current Haswell Xeon E3-1200 lineup for the desktop:

Desktop Haswell Xeons

Notice how the 1270 runs at 3.5 GHz and costs $300, while the 1280 runs just an extra 0.1 GHz faster but costs $600, a whopping 100% price bump for a 3% perpformance increase.

Is there something so much better about the E3-1280 that I'm not seeing here? Or is Intel just trying to take advantage of people who like having the top of the line, regardless of how small the margin to the "top" might be?
 

elemein

Honorable
Mar 4, 2013
802
0
11,160
- Extra 100 MHz
- Higher TDP (much more significant than one would think)
- Most likely much higher binning (also much more significant than one would think.)

So is it worth it to a consumer? Absolutely not.

But to a server host where heat is an issue and performance is key, the extra binning and higher TDP really does help. Even if just a bit, it's all about that "bit."
 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
Ah, very interesting.

I'm confused though, I thought that TDP was equivalent to heat output, meaning that the 82W on the 1280 would be *worse* for a server host than the 80W on the 1270.

Also, I'm new to this hardware stuff in general, what is this "binding" thing of which you speak?
 

elemein

Honorable
Mar 4, 2013
802
0
11,160


A higher TDP would be worse for a situation where you have a heat issue, but consider the following situation.

CPU 1 and CPU 2 are based on the same architecture with the same performance; etc. etc.

CPU 1 has higher binning and a TDP of 90W.
CPU 2 has lower binning but a TDP of 90W too.

Logic would dictate that the CPUs would produce the same amount of heat over an hour when pushed to maximum output.

The thing is, while they bot will produce a total (theoretical) heat output of 90W, CPU 1 can do so with less throttling and less voltage because it is higher binned; meaning that you get more performance and output per watt.

So binning is indeed important.

 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
Hmm, I'm still a bit fuzzy on this binning concept. So are the 1280 just 1270's that happened to have come out of the manufacturing processes with fewer defects, or are they intentionally made differently?

Is there anywhere in the Intel product charts where one can see just exactly how much better per watt the 1280 will perform over the 1270? I would imagine it's pretty substantial if the price increase is literally doubled.
 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
So, if I understand correctly, they do both come out of the same manufacturing batch, and then the 1270 or 1280 label gets put on after the fact, depending on an individual chips performance (which I gather is just a matter of luck).

However, that still doesn't explain why the 1280 is so much more expensive. I'd be very curious to see some real numbers about how much better it performs over the 1270, and in what way.
 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
You have all yet to answer what *exactly* it is that justifies a price doubling from $300 to $600. What *exactly* is this "efficiency" or "performance" increase that you speak of, and what are the numbers?
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


There is no exact justification. Yield is somewhere in the equation but the rest is whatever Intel thinks they can get away with while keeping an average of 60% margin. It's not a linear scale.

$215 Xeon E3-1230 3.2/3.6
$250 Xeon E3-1240 3.3/3.7 +45
$328 Xeon E3-1270 3.4/3.8 +78
$612 Xeon E3-1280 3.5/3.9 +284
$885 Xeon E3-1290 3.6/4.0 +273

 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
Ah, so then it would indeed be futile to search the 1280's specs, benchmarks, or reliability characteristics for anything that seems doubled with respect to the 1270? The only thing that's even close to doubled is the price?

Do I have that straight?
 

MajinCry

Distinguished
Dec 8, 2011
958
0
19,010
Welcome to corporations! Doing anything they can to get as much money from you as possible.

AMD isn't innocent, as that company has done the same thing. However, intel makes it FAR more apparent, as you have shown. Don't forget the extreme line o' proccies that have only a small boost over their non extreme counterparts, yet cost several times more.
 

jed

Distinguished
May 21, 2004
314
0
18,780


Yes I see your point, but both company's do this, AMD FX 8350 overclocked with nothing extra
to 5Ghz and call it the FX 9590 for a 700.00 dollar premium, is just totally insane.

 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
I don't understand how they get away with doing that in the consumer market, or for that matter any any other market. Even a person or organization with outrageous amounts of money wouldn't be justified paying a premium that is so grossly over-marked.

It's not even about the money specifically, it's about common decency and respectful business ethics.

Let's take an analogy. Imagine that you're obscenely rich. One day you happen to to get in a cab. You tell the cabby where you want to go, and he responds "That'll take 35 minutes if you pay me $40, or 34 minutes if you pay me $80." Even if you're the sort of person who lights his evening cigars with $100's every day, you probably still wouldn't give him the $80, because your gut instinct would tell you that such a blatantly horrible deal is just insulting.
 

elemein

Honorable
Mar 4, 2013
802
0
11,160


You're thinking about it all wrong. It is not only a good business practice, but it is fully respectable.

First and foremost, this isnt the consumer market. Xeons arent for general consumers; consumers CAN buy them, but they aren't meant for them. They are for professionals; servers, workstations, renderers; etc. They are for professionals.

Secondly, you're not paying "for an extra minute off the cab ride", that analogy isnt right. Something like this would be more accurate:

You own a lawyer firm and are hiring a new lawyer. One lawyer has full honours and graduated second in his class. While another lawyer was in the exact same school in the exact same class, but is first.

Of course, both are amazing choices, but it's your choice if you really want the best of the best.

The extra 300$ is going to extra binning because the more expensive Xeon is the best of the best out of the Xeon batches. It has achieved a near perfect level of doping and atom distribution. It is the best of the best.

Will the lawyer second in his class do well? Yes, of course. But he isnt the best of the best.
 

rutski89

Honorable
Aug 11, 2013
28
0
10,530
If the difference between the two lawyers is that one of them averaged 99.5/100 on his exams, and the other averaged 99.8/100 on his exams, then I would say that the second lawyers has no standing to charge literally double what the first charges. The first may be able to get away with a slight price increase just because he's formally tagged at being at the top, but he certainly can't charge double. Your analogy does not hold.

Also, I'm still new to all this hardware stuff, so here are some innocent questions, if someone would be kind enough to educate me:

A) What exactly is doping, and what benefit comes from having a near near perfect level?
B) Likewise, what is atom distribution? And likewise what benefit comes from having a near near perfect distribution?
C) Do said benefits actually warrant a price increase of 100%?
 

TRENDING THREADS