• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Clarkdale+IGP tested

Nice find. I wouldn't say the IGP perfomance is bad, looking at the 1080p playback. Really, IGPs aren't meant for anything 3D, IMO.

I noticed the test was run on an H55 board, and wonder if it's one of those boards that sport built-in SSD. That'd be contributing to the overall performance to an extent. (which isn't a bad thing at all)

As long as it handles Aero, accelerates 1080p (including audio support), and runs a 30" monitor, it looks like Intel has another winner on its hand, waiting.
 
Google English Translated for those english-speakers interested in attempting to read the article.

edit: Interesting TDP numbers in the table on this page...bumping the GPU clockspeed up from 733MHz to 900MHz (23% increase) requires the socket TDP to increase from 73W to 87W (20% increase).

I would have expected better clock scaling vs. TDP results for a 45nm HKMG process tech IC (Vcc for the IGP must be markedly increasing at that point on the shmoo), but I guess the real question here is if 733MHz IGP clockspeed is "good enough" then who cares if those "good enough" fps are improved by a mere 20% on the i5-661? I don't really get what this SKU is targeted to accomplish that an i5-660 would achieve with less power-consumption.

edit 2: And in looking at the clock-for-clock comparisons between clarkdale and havendale benches in the tables on this page I have to say I am kinda astounded at the results indicating there are basically zero IPC improvements done to Westmere core architecture over that of Nehalem. IPC is pretty much dead-on the same.

Meaning westmere is about as close as it comes to doing a true "dummy shrink", just as Inteluser has been telling us for the past 6-9months.

Power consumption numbers for identical clockspeeds are on this page. At idle 2.4 GHz clarkdale improves power consumption from 62W to 54W (13% lower) and at load it improves from 102W to 91W (12% lower). At 3.067GHz idle power remains the same (naturally, cores are turned off regardless of clockspeed via power-gate) but at full load power consumption goes to 102W. Not exactly a power-sipper given that this is a dual-core (not quad) 32nm cpu.

edit 3: This quote is FTW :thumbsup:

it is a pity that, in the face of NVIDIA GeForce 9400 IGP chipset, Clarkdale graphics core as the Qing like knives against the Western cannon, completely annihilated

http://64.233.169.132/translat...VVvzxpY9cAINcejgA#view
Click to expand...
 
Im with lopri on this, Intel doesn't need a performance part for this market. As long as it does h.264 and aero perfectly, it has hit its design goals. Gamers aren't going intel graphics (until larrabee) and non-gamers don' care.
 
Originally posted by: ilkhan
Im with lopri on this, Intel doesn't need a performance part for this market. As long as it does h.264 and aero perfectly, it has hit its design goals. Gamers aren't going intel graphics (until larrabee) and non-gamers don' care.
Click to expand...

That!



I would be nice if the Intel IGP had the same perf as a a 780G IGP (Q1 2008-Q1 2010, 2 years gap, I still remember the 10X slides) but not a deal braker if they don't! (for the specific market)

 
Originally posted by: Idontcare
Google English Translated for those english-speakers interested in attempting to read the article.

edit: Interesting TDP numbers in the table on this page...bumping the GPU clockspeed up from 733MHz to 900MHz (23% increase) requires the socket TDP to increase from 73W to 87W (20% increase).

I would have expected better clock scaling vs. TDP results for a 45nm HKMG process tech IC (Vcc for the IGP must be markedly increasing at that point on the shmoo), but I guess the real question here is if 733MHz IGP clockspeed is "good enough" then who cares if those "good enough" fps are improved by a mere 20% on the i5-661? I don't really get what this SKU is targeted to accomplish that an i5-660 would achieve with less power-consumption.

edit 2: And in looking at the clock-for-clock comparisons between clarkdale and havendale benches in the tables on this page I have to say I am kinda astounded at the results indicating there are basically zero IPC improvements done to Westmere core architecture over that of Nehalem. IPC is pretty much dead-on the same.

Meaning westmere is about as close as it comes to doing a true "dummy shrink", just as Inteluser has been telling us for the past 6-9months.

Power consumption numbers for identical clockspeeds are on this page. At idle 2.4 GHz clarkdale improves power consumption from 62W to 54W (13% lower) and at load it improves from 102W to 91W (12% lower). At 3.067GHz idle power remains the same (naturally, cores are turned off regardless of clockspeed via power-gate) but at full load power consumption goes to 102W. Not exactly a power-sipper given that this is a dual-core (not quad) 32nm cpu.

edit 3: This quote is FTW :thumbsup:

it is a pity that, in the face of NVIDIA GeForce 9400 IGP chipset, Clarkdale graphics core as the Qing like knives against the Western cannon, completely annihilated

http://64.233.169.132/translat...VVvzxpY9cAINcejgA#view
Click to expand...
Click to expand...

Really ! Lets run these in same test and see if that holds true. I understand what your saying clock for clock . Don't rush to judgement .

Dual-core desktop processors under the Clarkdale family also won't ship until early 2010 but should be some of the first Intel chips built on a smaller, more efficient 32 nanometer process. Most will have Hyperthreading and should be topped by 3.2GHz, 3.33GHz and 3.46GHz Core i5 models that ramp up to 3.46GHz, 3.6GHz and 3.73GHz using Turbo Boost. Two lower-clocked but functionally identical Core i3 chips at 2.93GHz and 3.06GHz are also due, as is a lone 2.8GHz Pentium that would have Hyperthreading disabled and a drop to 3MB of Level 2 cache from the 4MB present on the other dual-core systems.

I haven't seen were intel is offerring a 2.4 clarksdale. You see were picture is tainted.

Getting 3.73+ from Clarkdale out of box from Dell is huge . Being able to do 4 threads at once is bonus. Getting 5ghz O/c is walk in park.

The 3.o6 numbers look really good considering turbo was off. Weather people like it or not Turbo works and its a feature . People that but A pc from Dell will rave over these things and we all know it. 3.73 single threaded performance .+ if intel holds true to the $500 exstreme dual core unlocked. That would lock up the middle for intel . The high end is intels already and the low end . The unlocked i5 dual core is the final blow. Thats what brother in law says .

 
Originally posted by: jones377
http://global.hkepc.com/3878/page/1#view

Great CPU performance, still terrible IGP performance where the difference is the smallest in 3Dmark (who would have thought?)
Click to expand...


I found the IGP was rather good it was running @ 700mgz not 733mgz The highend runs @ 900mgz So this will be as good as the NV solution shown .

Also the I3 comes with 3.o6 I3 is intels low end . $86.

 
Originally posted by: Nemesis 1
Originally posted by: Idontcare
Google English Translated for those english-speakers interested in attempting to read the article.

edit: Interesting TDP numbers in the table on this page...bumping the GPU clockspeed up from 733MHz to 900MHz (23% increase) requires the socket TDP to increase from 73W to 87W (20% increase).

I would have expected better clock scaling vs. TDP results for a 45nm HKMG process tech IC (Vcc for the IGP must be markedly increasing at that point on the shmoo), but I guess the real question here is if 733MHz IGP clockspeed is "good enough" then who cares if those "good enough" fps are improved by a mere 20% on the i5-661? I don't really get what this SKU is targeted to accomplish that an i5-660 would achieve with less power-consumption.

edit 2: And in looking at the clock-for-clock comparisons between clarkdale and havendale benches in the tables on this page I have to say I am kinda astounded at the results indicating there are basically zero IPC improvements done to Westmere core architecture over that of Nehalem. IPC is pretty much dead-on the same.

Meaning westmere is about as close as it comes to doing a true "dummy shrink", just as Inteluser has been telling us for the past 6-9months.

Power consumption numbers for identical clockspeeds are on this page. At idle 2.4 GHz clarkdale improves power consumption from 62W to 54W (13% lower) and at load it improves from 102W to 91W (12% lower). At 3.067GHz idle power remains the same (naturally, cores are turned off regardless of clockspeed via power-gate) but at full load power consumption goes to 102W. Not exactly a power-sipper given that this is a dual-core (not quad) 32nm cpu.

edit 3: This quote is FTW :thumbsup:

it is a pity that, in the face of NVIDIA GeForce 9400 IGP chipset, Clarkdale graphics core as the Qing like knives against the Western cannon, completely annihilated

http://64.233.169.132/translat...VVvzxpY9cAINcejgA#view
Click to expand...
Click to expand...

Really ! Lets run these in same test and see if that holds true. I understand what your saying clock for clock . Don't rush to judgement .

Dual-core desktop processors under the Clarkdale family also won't ship until early 2010 but should be some of the first Intel chips built on a smaller, more efficient 32 nanometer process. Most will have Hyperthreading and should be topped by 3.2GHz, 3.33GHz and 3.46GHz Core i5 models that ramp up to 3.46GHz, 3.6GHz and 3.73GHz using Turbo Boost. Two lower-clocked but functionally identical Core i3 chips at 2.93GHz and 3.06GHz are also due, as is a lone 2.8GHz Pentium that would have Hyperthreading disabled and a drop to 3MB of Level 2 cache from the 4MB present on the other dual-core systems.

I haven't seen were intel is offerring a 2.4 clarksdale. You see were picture is tainted.

Getting 3.73+ from Clarkdale out of box from Dell is huge . Being able to do 4 threads at once is bonus. Getting 5ghz O/c is walk in park.

The 3.o6 numbers look really good considering turbo was off. Weather people like it or not Turbo works and its a feature . People that but A pc from Dell will rave over these things and we all know it. 3.73 single threaded performance .+ if intel holds true to the $500 exstreme dual core unlocked. That would lock up the middle for intel . The high end is intels already and the low end . The unlocked i5 dual core is the final blow. Thats what brother in law says .
Click to expand...

Nemesis did you actually check out any of the links embedded in my quotes? My post is speaking to the data presented in the links, if the data are in error then go ahead and say so, but I don't think I mistated any conclusions given the data.
 
No data looks good . But IGP was running @ 700mhz . There was no turbo . Everone and there mother can downplay What Clarkdale brings. The proof will be 1st qt profits report from intel . This is a great chip . it will be intels $$$$$$$$$$ maker. Thats the bottom line. Forum people can argue till there blue in the face about performance. Intel is going to offer it cheap and at very high level to the WORLD not US. THE world will eat these things like candy. INTEL 1st QT results is what I am talking about and its what your all afraid of.
 
If i remember the 4500HD has only 2 ROPs!

If Intel didn't increased the ROPs (i only saw the 6 vs 5 regarding the SPs) the performance improvements they made at 700MHz (-100MHz) are very good!
(although meaningless, the 2007-2008 games are still not playable not even at 1024x768 lowest set.)

The pentium IGP is failure (533MHz without H264 HD accel?.), but the price is very good!


I don't get the 14W increase of the 661 (why intel to do this? what perf advantage 900MHz will bring to 733MHz, it's ridiculous)

The i3-530 (123$) has a killer price!

You can see that in some apl. like H264 enc (2p)
the clarkdale arch. are only -25% in relation with a quad phen ii (same clock)

 
Ya the pentium IGP is bad. But Intel still has old stock to clear out . This Cpu 6 months after release will be $40 or less. Let intel ramp up 32nm. These are simply first releases.

Ya Intels IGP is just fine, comparing in gaming is laughable none do well at all . SO the argument is a red Herring. Nothing more. This is solid chip for intended market. Your right about the 900mgz increases considering gaming is a joke on all IGPs But it will show improvements in benchies . The cost in effiency is outragious right.
 
Probably the scenario regarding the pentiums is:

Q1 2010 2,8GHz 87$
Q3 2010 2,8GHz 77$
Q1 2011 2,8GHz 67$

If you remember, we had in Q4 2007/Q1 2008 45nm cores but in order the architecture to go down to 53$/43$ (3300/3200) it took 1,5 year or more! (Q3 2009)

So I think the E3X00 arch. will stay in the market for a while!

I don't think it will take 1,5+ years like 45nm core, to go down to 56$ (former 53$), but something like 1 year (1 of the reasons: they don't like NV :laugh:

EDIT*

Actually i changed my mind regarding the life time of the E3X00, i think that your scenario is better (6 months)!

In the 1,5 year cycle of 45nm the socket was the same!
With clarks the sock is changing and there is a possibility in Q3 2010, intel to have new low cost m/b chipsets!
Also since in Q3 2010 the chipsets will be 45nm and Intel will probably be completely alone in the chipset business for neh, intel will need more 45nm space, so off you go E3X00!
So (depending also on the stock situation of the E3X00) the possibilities for Q3 2010 "Clarkdales celerons" is good!

 
The Core i5-670 might just entice me to upgrade my E6850. It's a nice dual-core with a high clock speed and a reasonable price. Plus it?s basically guaranteed to overclock to 3.73GHz thanks to officially sanctioned turbo boost

This is what I've been looking for in a CPU, unlike the horrifically overpriced quad-core i7s that aren?t much faster in games because of low clock speeds, and because most games can?t use more than two cores.

Quad-core is vastly overrated, as is the i7 platform.
 
Originally posted by: MODEL3
If i remember the 4500HD has only 2 ROPs!

If Intel didn't increased the ROPs (i only saw the 6 vs 5 regarding the SPs) the performance improvements they made at 700MHz (-100MHz) are very good!
(although meaningless, the 2007-2008 games are still not playable not even at 1024x768 lowest set.)

The pentium IGP is failure (533MHz without H264 HD accel?.), but the price is very good!


I don't get the 14W increase of the 661 (why intel to do this? what perf advantage 900MHz will bring to 733MHz, it's ridiculous)

The i3-530 (123$) has a killer price!

You can see that in some apl. like H264 enc (2p)
the clarkdale arch. are only -25% in relation with a quad phen ii (same clock)
Click to expand...

I HIGHLY doubt the IGP is that uncompetitive as the article says. I mean first of all, they are using Win 7 drivers for Vista 32 bit FFS.

And also take a look at this one: http://global.hkepc.com/1865/page/11#view

Notice here how using DDR2-800 the G45 is only 2.5x behind Geforce 9400 in the Crysis benchmark. How it goes from that to a 6x behind(as per Clarkdale benchmark) with DDR3-1333 memory is beyond me. And for numbers only, my X3000 system with E6600 gets better numbers in Crysis 1024x768 Low than HKEPC does with the X4500(2x better actually). The drivers are seriously messed up there...

Power consumption numbers for identical clockspeeds are on this page. At idle 2.4 GHz clarkdale improves power consumption from 62W to 54W (13% lower) and at load it improves from 102W to 91W (12% lower). At 3.067GHz idle power remains the same (naturally, cores are turned off regardless of clockspeed via power-gate) but at full load power consumption goes to 102W. Not exactly a power-sipper given that this is a dual-core (not quad) 32nm cpu.
Click to expand...

IDC, you kinda realize that the power consumption figures are system level right? Compare for AT's numbers, that's actually really good: http://www.anandtech.com/mb/showdoc.aspx?i=3430&p=10

And the real numbers to compare is with the E8500. I'd say 14W in idle and 16W in load is substantial!

I'm kinda expecting to be "surprised" with a secret with the 87W Clarkdale. Considering how much better the 73W is compared to the E8500, 87W would just be on par, but ignoring power consumption figures, there might be more than the clock speed increase. Maybe it has a "Graphics Turbo Mode". 1100MHz IGP please!!

Meaning westmere is about as close as it comes to doing a true "dummy shrink", just as Inteluser has been telling us for the past 6-9months.
Click to expand...

Regarding Westmere and "cache improvements" I have seen on Intel presentations. I have concluded sometime ago that it might only apply to Westmere-EX.

I noticed the test was run on an H55 board, and wonder if it's one of those boards that sport built-in SSD. That'd be contributing to the overall performance to an extent. (which isn't a bad thing at all)
Click to expand...

I hear "Braidwood", which is the codename for the acceleration technology is dead. At least in P55 chipsets. The test says its running X25-M SSD however. No, the improvements are all due to the CPU. I find its a bit less than on Bloomfield, probably thanks to (much)less bandwidth. Of course you don't have to believe me though.
 
Originally posted by: IntelUser2000
Originally posted by: MODEL3
If i remember the 4500HD has only 2 ROPs!

If Intel didn't increased the ROPs (i only saw the 6 vs 5 regarding the SPs) the performance improvements they made at 700MHz (-100MHz) are very good!
(although meaningless, the 2007-2008 games are still not playable not even at 1024x768 lowest set.)

The pentium IGP is failure (533MHz without H264 HD accel?.), but the price is very good!


I don't get the 14W increase of the 661 (why intel to do this? what perf advantage 900MHz will bring to 733MHz, it's ridiculous)

The i3-530 (123$) has a killer price!

You can see that in some apl. like H264 enc (2p)
the clarkdale arch. are only -25% in relation with a quad phen ii (same clock)
Click to expand...

I HIGHLY doubt the IGP is that uncompetitive as the article says. I mean first of all, they are using Win 7 drivers for Vista 32 bit FFS.

And also take a look at this one: http://global.hkepc.com/1865/page/11#view

Notice here how using DDR2-800 the G45 is only 2.5x behind Geforce 9400 in the Crysis benchmark. How it goes from that to a 6x behind(as per Clarkdale benchmark) with DDR3-1333 memory is beyond me. And for numbers only, my X3000 system with E6600 gets better numbers in Crysis 1024x768 Low than HKEPC does with the X4500(2x better actually). The drivers are seriously messed up there...
Click to expand...

Maybe it is just the drivers like you are saying!

In any case it is highly unlikely to be competitive with the Q1 2008 780G IGP in games!

I really don't care, as long as it does good 2D and HD playback!

We will see at launch in Anand's review the real perf.!

 
http://www.tomshardware.com/re...g-chipset,2381-11.html

It's really not that far away. The drivers have come seriously a long way in the performance/quality/compatibility department. So the Clarkdale IGP is 2-2.5x faster? That'll allow it to be on par with 780G, which isn't the fastest anymore. Add some final drivers, probably the top end current IGP.

I get 6-7 fps in Crysis with the same settings HKEPC got with the G45 got 3.9 fps in. Scale that with final hardware and the improvements with Clarkdale.

The official release is 3+ months away. Considering the September 8th release date for Lynnfield and how scant the information is still, lot can change in that time.
 
Regarding the 661, I think probably it is just an experiment (like what they are doing with the Pilot Production of E6500)
Certainly, Intel right now is very strong and can afford such experiments!

But if AMD play its cards right (marketing PR against 661 types of processors) the 661 concept can lead to more troubles than gains!

I guess for 2D applications and HD playback the 733MHz cores can do the job just fine!
(unless they don't, so all the range will be a failure, but i doubt that this is the case, i think that for these tasks the 733MHz cores will be just fine after 1 or 2 months that the drivers and the applications will be a little more mature!) (if not at launch!)

So the difference essentially between the 733MHz cores and the 900MHz cores will be like 20-25% in the games/synthetic benchmarks & and in some GPGPU enabled applications!

All the above differencies for me are meaningles!

Yes, Intel can promote a higher 3D Mark Vantage score for example, but at the same time for applications like the above (games/synthetic benchmarks & and in some GPGPU enabled applications) even a Q1 2008 IGP like the 780G is better!

And you lose some things like VT-d & TXT (of cource i don't think that all the people will care for these loses)
and you go to 87W which is not exactly that far from some 95W quad Phenom IIs
(although the actual difference will be much bigger than 8W) and don't forget that we will have some new cheap AMD quad CPUs like 6XX series at 65W and also some higher priced ones like 6XXe series at 45W.

If AMD play the marketing cards right (just one example: if AMD focus on the "best" 900MHz types of CPUs and try to correlate why a Quad PhenomII /Athlon II + 785G chipset (or a feature Q1 2010 chipset)
is better than even a high IGP clocked "661 type of CPU"
then it can succeed to discredit the Intel effort with 661! (there are other examples also)

What I mean is, that i can see the marketing possibilities for Intel regarding the 661, but i can also see the marketing possibilities for AMD!

Unless the IGP core in 661 is not the same with the other series IGP core!

 
You must log in or register to reply here.

TRENDING THREADS

Back
Top