Intel has four-core Opteron stuffer set for August

[coffeetime]

http://www.channelregister.co.uk/2007/06/29/intel_2ghz_clovertownlv/

Exclusive Intel looks set to blunt AMD's August Barcelona processor release by handing customers a 2.0GHz version of its four-core Clovertown, The Register has learned.

Just this morning, AMD announced that in August it will ship customers 1.9GHz low power and 2.0GHz standard editions of Barcelona - a four-core version of the Opteron server chip. Servers with the new chips should arrive in bulk in September and October. Intel, however, isn't sitting still from what we hear and looks set to effectively one-up AMD.
Click here to find out more!

Chipzilla also in August will release a low voltage version of its four-core Xeon code-named Clovertown that runs at 2.0GHz, while consuming just 50W max power. By contrast, AMD's upcoming standard edition Opteron will eat up 95W, while the low power chip will eat up 68W. Intel's new 2.0GHz chip will also boast a 1333MHz front side bus, up from today's 1066MHz FSBs, according to our sources.

Intel currently ships a 50W version of Clovertown running at 1.86GHz, although most of the parts consume 80W.

The release of the 2.0GHz chip proves a nasty stab at AMD, which hoped to have the August fanfare to itself.

AMD looked to best Intel out of the gate with the new Opterons, although that seems like a bit of a dream now. The chip maker told us yesterday that its 2.0GHz four-core chip will beat today's Opterons by 50 per cent on average and will "outperform the competition on certain workloads."

Such talk is a far cry from AMD's previous chest-thumping around beating Clovertown by 40 to 50 per cent on a wide range on benchmarks. It seems AMD will need faster versions of Opteron, due out in the fourth quarter, to do such damage.

Customers must love the ferocity with which Intel and AMD are going at each other right now. AMD hands you twice the cores and then Intel slaps its rival in the face by handing you a part that matches AMD's clock rate while consuming less power.

[josty2]

And still the motherboard using 1333FSB will eat a lot of energy. Combining that with the wonderfull TDP calculation they have, means... That it will probably will be a very close call, considering AMD thinks their IPC on K10 will be higher than Clovertowns...
Interesting to see who´ll win in august!

[eva2000]

Intel must be having lots of fun kicking AMD while they're down LOL

[alfaunits]

Mobo and FBDIMM sure make the LV look useless.

[Shintai]

Mobo and FBDIMM sure make the LV look useless.

Not really. The mobo dont use much more than the nVidia 3000 series solution. And 4 FB-DIMMs add what...15W?
And the bigger DIMMs, the less amount is taken from the AMB chip of the overall memory usage.

And in these systems you would have 2 CPUs, 8 cores. So the 50W vs 68W is just twice as bad. And the FB-DIMM offset is the same. Singlesockets use standard DDR2.

[alfaunits]

How much does an octal-core Opty box use? (according to some news here < 250W)
How much does a quad-core Woodcrest use? (>300W, MINIMUM, with LV 5148).
Opty has a clear advantage here. But that hardly interests anyone here (not me for sure).

[Shintai]

How much does an octal-core Opty box use? (according to some news here < 250W)
How much does a quad-core Woodcrest use? (>300W, MINIMUM, with LV 5148).
Opty has a clear advantage here. But that hardly interests anyone here (not me for sure).

Under 250W for 4 dualcore Opterons? Or under 250W for 2 quadcore K10?

And 300W+ for the dual quadcore xeon? So 2x50W CPUs and 200W what? Maybe if it had a R600 card in it.

As a person that got PLENTY of dual quadcores and 2, 4 and 8way opterons its nothing but marketing BS when I use an actual load checker on the AC side from normal bought OEM servers.

Unless you refer to this:
http://vd.verysell.ru/files/ie/252_10_DOCUMENT_Server_Section_PUBLIC.ppt
Where they add 11% here and there extra and use numbers that are only theorectical max. They even claim 2.66 and lower Woodcrests uses 72W and not 65W. Yet not even 2.93Ghz C2Ds can get there in tests.

Its one big marketing FUD file. Tho its funny enough also kills the IPC dreams on K10.

[Movieman]

8 core clovertown at 3157=450w measured at 100% load
1-ATI X550 vid, one 80 gig drive
PCP+C 610w silencer

[[XC] hipno650]

8 core clovertown at 3157=450w measured at 100% load
1-ATI X550 vid, one 80 gig drive
PCP+C 610w silencer

was that out of the wall? because factor in the PSU efficiency and your component consumption is less. the best part is that is 450watt's of XS WCG power!!!

[alfaunits]

Under 250W for 4 dualcore Opterons? Or under 250W for 2 quadcore K10?
Both. Not saying _I_ measured it! (it was an article posted here with an AC thingie).

And 300W+ for the dual quadcore xeon?
No. For dual 5148, with 4 Raptors and onboard GPU. (aside from -4 drives, the LEAST you can go with Woodcrest power wise). This one I measured with a UPS.

As a person that got PLENTY of dual quadcores and 2, 4 and 8way opterons its nothing but marketing BS when I use an actual load checker on the AC side from normal bought OEM servers.
You mean the AC side checker is s***, or it's the only right way?
I'm not arguing performance/watt, or performance at all of Optys, but at same speeds and cores, Optys do consume less.

[Shintai]

You cant use his 450W for much. I would assume that besides the CPUs, also the chipset and memory got a slight OC? It could be nice with stock speed measurement tho :D

[Shintai]

Both. Not saying _I_ measured it! (it was an article posted here with an AC thingie).


No. For dual 5148, with 4 Raptors and onboard GPU. (aside from -4 drives, the LEAST you can go with Woodcrest power wise). This one I measured with a UPS.


You mean the AC side checker is s***, or it's the only right way?
I'm not arguing performance/watt, or performance at all of Optys, but at same speeds and cores, Optys do consume less.

On the AC side, both Dell 1950s and Dell 2950s here with dual 2.33Ghz quadcores (80W), 4GB in 4 FB-DIMMs and either 1 disk (rendernodes) or up to 6 disks, either 500GB disks or 300GB SAS with RAID controller do NOT go over 280W on the AC side on the heaviest one. If I used 2.66 or 3Ghz Quads, sure.

Dont let fancy marketing fool you. Reality is quite different. Even a dual dualcore Opty box with 95W CPUs uses more.

We even had Sun out with a testserver aswell. After all their BS about powerconsumption. Only to go away with red embarrased faces when the wattage was measured against the Dell servers on the AC side!

[Movieman]

hipno650;2286309']was that out of the wall? because factor in the PSU efficiency and your component consumption is less. the best part is that is 450watt's of XS WCG power!!!

Yes, out of the wall.
To answer the other question:
chipset is running normal voltage as are the chips(1.35v showing in coretemp and 1.28v showing in bios)
Memory is on stock voltage with slight OC to 350mhz

[Syn.]

Chipzilla also in August will release a low voltage version of its four-core Xeon code-named Clovertown that runs at 2.0GHz, while consuming just 50W max power. By contrast, AMD's upcoming standard edition Opteron will eat up 95W, while the low power chip will eat up 68W. Intel's new 2.0GHz chip will also boast a 1333MHz front side bus, up from today's 1066MHz FSBs, according to our sources.

Intel currently ships a 50W version of Clovertown running at 1.86GHz, although most of the parts consume 80W...

Something that needs to cleared up when ever people use the TDP calculations.

Intel uses Typical TDP(Thermal Design Power)
AMD uses Maximum TDP(Thermal Design Power)

As they correct them self's afterwords by saying that a Clovertown with 50W TDP usually ends up consuming 80W. Intel is yet again using dirty marketing tricks to full the consumer.

Intel might release a 2Ghz Clovertwon but we still don't know what 2Ghz of Barcelona is.

[accord99]

Something that needs to cleared up when ever people use the TDP calculations.

Intel uses Typical TDP(Thermal Design Power)
AMD uses Maximum TDP(Thermal Design Power)

Except, it has been show again and again in CPU power consumption tests that this generalization is inaccurate with Core-based processors.

[Poodle]

In real life when it comes to what the whole system pulls AMD will probably be lower power though as they tend to scale better with CnQ etc and Barcelona has new such feats also.

On full load Intel will probably be the winner.

[Shintai]

Something that needs to cleared up when ever people use the TDP calculations.

Intel uses Typical TDP(Thermal Design Power)
AMD uses Maximum TDP(Thermal Design Power)

As they correct them self's afterwords by saying that a Clovertown with 50W TDP usually ends up consuming 80W. Intel is yet again using dirty marketing tricks to full the consumer.

LOL, sure...in what world?

http://www.xbitlabs.com/articles/cpu/display/core2duo-shootout_11.html#sect0
http://www.xbitlabs.com/articles/cpu/display/core2extreme-qx6700_11.html#sect0

And thats measured before the VRM.

Perhaps you should stop the FUD when you dont know what you are talking about.

[alfaunits]

Dont let fancy marketing fool you. Reality is quite different. Even a dual dualcore Opty box with 95W CPUs uses more.

Seriously, do I look like an AMD fanboy? I've had nothing but bad experience with it, but I give credit where credit's due.
http://www.anandtech.com/showdoc.aspx?i=2892&p=10
Your loss if you continue to assume/assert/argue Opty uses more.

[savantu]

...

Intel uses Typical TDP(Thermal Design Power)
AMD uses Maximum TDP(Thermal Design Power)

...

For your information K10 will use Intel's system.Why ? Because it is better , the way in which AMD calculates TDP ( maxVcc*maxIcc) is absurd and can't be reached in real world.

[Shintai]

Seriously, do I look like an AMD fanboy? I've had nothing but bad experience with it, but I give credit where credit's due.
http://www.anandtech.com/showdoc.aspx?i=2892&p=10
Your loss if you continue to assume/assert/argue Opty uses more.

Ye, now you just need to learn and compare apples to apples in terms of performance. I can take a X2 3600 too and compare it with a Q6600 and show how efficient the X2 3600 is. And how the Q6600 is a powerhog.

So look abit on these pages instead:
http://www.anandtech.com/showdoc.aspx?i=2892&p=11
http://www.anandtech.com/showdoc.aspx?i=2892&p=7

[accord99]

Seriously, do I look like an AMD fanboy? I've had nothing but bad experience with it, but I give credit where credit's due.
http://www.anandtech.com/showdoc.aspx?i=2892&p=10
Your loss if you continue to assume/assert/argue Opty uses more.

Server system power usage varies a lot, there are reviews of Woodcrest systems with low power usage like here:

http://www.2cpu.com/review.php?id=114&page=15
http://www.2cpu.com/review.php?id=112&page=8
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2793&p=5
http://www.techreport.com/etc/2006q2/woodcrest/index.x?pg=9
http://www.tomshardware.com/2006/10/26/intel_woodcrest_and_amd_opteron_battle_head_to_hea d/page9.html

Interestingly, Anand's first Woodcrest system had lower power usage than the Opteron 285 system, while the AMD provided Woodcrest system had higher power consumption than the 285.

[BitpowerPM]

How much does a quad-core Woodcrest use? (>300W, MINIMUM, with LV 5148).

My two Xeon E5320 @ L5320 with 4x1 GB FB-DIMM, 7800GTX and 6x S-ATA HDD (Seasonic S12-600) consumes no more than 270 watts under cpu full-load (Folding).
When using a less power-hungry graphics-card you go under 250 watts with a powerful eight-core system easily.
Please show me an Opteron system @ equal performance in threaded applications or virtualizing software at an equal power-consumption. :rolleyes:

[alfaunits]

My two Xeon E5320 @ L5320 with 4x1 GB FB-DIMM, 7800GTX and 6x S-ATA HDD (Seasonic S12-600) consumes no more than 270 watts under cpu full-load (Folding).
OK, so what is eating 80+W on my system vs. yours? )or are you measuring different wattage - i.e. not from a UPS)

Please show me an Opteron system @ equal performance in threaded applications or virtualizing software at an equal power-consumption. :rolleyes:
Never said Opty's faster.

@Shintai, I have no clue what you meant with your links, for they sure don't go in your favor.

[Donnie27]

Mobo and FBDIMM sure make the LV look useless.

Want to bet?

http://techreport.com/reviews/2007q2/intel-v8/index.x?pg=13

By finishing the job very quickly, the Xeon X5365 uses relatively little energy to render the scene. That's the V8 system's power efficiency ace in the hole. You've got to keep all of those cores busy in order for it to be power efficient, but it certainly can be.

RAM used.
Samsung ECC DDR2-667 FB-DIMM at 667MHz.

The newer processors have better Idle and Sleep low power states. They'll only use ECC DDR2 and then 3 as well:)

[Shintai]

@Shintai, I have no clue what you meant with your links, for they sure don't go in your favor.


Maybe you should check again. Performance/watt? Else you could just as easily compare a Pentium III with a 8way Opteron box and say the Opteron box suck and use more power. And you ahev been proven wrong several times now with several links.