What if Intel sells its future desktop microprocessors, ONLY with an IGP included?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: ShawnD1
Originally posted by: alyarb
heh looks like you'll have to take the i/o shield off that 7950. can't blame intel though. i've had OEM PCs with no expansion slots at all. my sister's dimension 4600 comes to mind (a single 32bit PCI with no "opening" for i/o).

I don't blame Intel for Compaq's screw ups. I'm just stating that this possibility of Intel having total video card dominance in Intel systems is already true, but for different reasons. There's nothing in the chipset that says an AMD or Nvidia card won't work, but the cards simply do not work in most OEM computers. My friend's Compaq has the slot design problem, your sister's computer has the same basic problem, my old emachines had a 120W PSU and it would black screen crash when I had a Radeon 7200 PCI plugged in. This is pathetic. If you need to upgrade an OEM computer's video card because you think the Intel Extreme graphics is too extreme, you need to buy a new case, a new PSU, and possibly a new motherboard. You basically need a whole new computer.

I know PC fans like myself tend to attack Apple computers for being throw away toys that can't be upgraded, but a standard OEM PC is the exact same.

Really no different from the integration/lack-of-expansion found on the laptop market. You get what you paid for, anything beyond that is gratis.
 

DrMrLordX

Lifer
Apr 27, 2000
23,183
13,270
136
Originally posted by: VirtualLarry
Just thinking about how Intel could conceivably corner the market on graphics. What if they ONLY released their newer CPUs with an IGP integrated, thus cutting AMD/ATI and NV out of the market totally?

Of course, there would be a "larrabee port" on the CPU, to support a larrabee co-proc. But none from AMD/ATI or NV, of course.

Just like Intel's splitting of the CPU socket into high-end and low-end, and choosing to use a proprietary system interface for both, that locks out competitors, what is to say that they won't finally do that with graphics too?

Aside from the fact that most OEM computers lack proper expansion slots (making your point somewhat moot as other posters have articlated), it should be noted that the market segment populated by boards with PCI-E slots as a common feature utilizes a standard expansion slot (PCI-E) for graphics. It's not like Intel can afford to cull PCI-E support from all their chips/chipsets due to the significant number of users that utilize PCI-E slots for cards other than graphics cards.

It's not like they can just axe PCI-E in favor of a proprietary slot intended only for Larrabee cards. There are too many buyers out there that want expansion slots that support a common industry standard (such as PCI-E). If Intel wants to continue selling to them (and they do), they're going to continue to support standard expansion slots.
 

exar333

Diamond Member
Feb 7, 2004
8,518
8
91
Wouldn't bother me if all CPU's came with an IGP; you don't have to use it, right? It would be really handy if your graphics card failed and you still wanted to use your computer.
 

alyarb

Platinum Member
Jan 25, 2009
2,425
0
76
of course, if the memory controller remains off-die, i will remain off-put.
 

jgigz

Senior member
Jul 14, 2006
413
0
76
Originally posted by: Denithor
How about removing the improperly used apostrophe from your title?

+1 for that as I'm reading the article during my grammar class.
 

Denithor

Diamond Member
Apr 11, 2004
6,298
23
81
Originally posted by: alyarb
of course, if the memory controller remains off-die, i will remain off-put.

Memory controller is on-die on these chips (incorporated into i7 architecture). :roll:


Originally posted by: jgigz
Originally posted by: Denithor
How about removing the improperly used apostrophe from your title?

+1 for that as I'm reading the article during my grammar class.

:D
 

VirtualLarry

No Lifer
Aug 25, 2001
56,587
10,226
126
Originally posted by: Denithor
Originally posted by: alyarb
of course, if the memory controller remains off-die, i will remain off-put.

Memory controller is on-die on these chips (incorporated into i7 architecture). :roll:
Except, it's not in the same die as the CPU, it's an MCM, connected by a QPI link.

Originally posted by: jgigz
Originally posted by: Denithor
How about removing the improperly used apostrophe from your title?

+1 for that as I'm reading the article during my grammar class.

Ok, ok, I'll fix it.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
Originally posted by: VirtualLarry
Originally posted by: Denithor
Originally posted by: alyarb
of course, if the memory controller remains off-die, i will remain off-put.

Memory controller is on-die on these chips (incorporated into i7 architecture). :roll:
Except, it's not in the same die as the CPU, it's an MCM, connected by a QPI link.

That's news to me! i7 is monolithic as far as I can tell.
 

deimos3428

Senior member
Mar 6, 2009
697
0
0
Originally posted by: Idontcare
Really no different from the integration/lack-of-expansion found on the laptop market. You get what you paid for, anything beyond that is gratis.
Somewhat of an improvement, isn't it, if you think about this with respect to future upgrades rather than with respect to introducing the on-chip IGP. Once you've got one, upgrading the integrated graphics on a low-end laptop will merely require a new CPU with a better IGP, instead of a new laptop.

[edit](In the case of desktops, that'd be a new motherboard vs. a new CPU...less of a value proposition, perhaps.)

 

VirtualLarry

No Lifer
Aug 25, 2001
56,587
10,226
126
Originally posted by: Phynaz
That's news to me! i7 is monolithic as far as I can tell.
There are no i7s with IGPs.

The Nehalem-based CPUs with IGPs, have a 32nm CPU MCM, connected by QPI, to a 45nm IGP, and the IMC is in the IGP portion, not the CPU portion.

 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: VirtualLarry
Originally posted by: Denithor
Originally posted by: alyarb
of course, if the memory controller remains off-die, i will remain off-put.

Memory controller is on-die on these chips (incorporated into i7 architecture). :roll:
Except, it's not in the same die as the CPU, it's an MCM, connected by a QPI link.

VirtualLarry is right, this is clarkdale/arrandale (32nm westmere) under discussion and for those products the memory controller no longer resides on the same piece of silicon as the cpu core and the L3$, Intel moved the memory controller and the PCIe controller to the MCM'ed graphic chips (which is still 45nm).

Gulftown will still have the memory controller be an IMC, but for clarkdale/arrandale the memory controller is not an IMC in the sense that we typically restrict the application/usage of the term.

I will be interesting to see what the damage is in moving the memory controller back off-die but still keeping it under the IHS.

Latency will obviously take a hit but as we've seen in a fair number of reviews lately the memory speed and latency doesn't seem to have much of an impact on desktop application performance. (presumably thanks in large part to the healthy caches that sit between the cpu cores and the memory controller now)

Servers are a whole other story of course, which is why gulftown retaining an IMC makes perfect sense versus what Intel is doing for the desktop segment.

Originally posted by: deimos3428
Originally posted by: Idontcare
Really no different from the integration/lack-of-expansion found on the laptop market. You get what you paid for, anything beyond that is gratis.
Somewhat of an improvement, isn't it, if you think about this with respect to future upgrades rather than with respect to introducing the on-chip IGP. Once you've got one, upgrading the integrated graphics on a low-end laptop will merely require a new CPU with a better IGP, instead of a new laptop.

[edit](In the case of desktops, that'd be a new motherboard vs. a new CPU...less of a value proposition, perhaps.)

True that, can't argue with you there.