JRockit-based SPECjbb2005 records achieved with Woodcrest

[TarTheDark]

"To summarize:
1. More than 50% better than the previous x86 results in the same niche
2. Almost 35% better than the latest Sun T2000 result
3. Compares favorably with much larger Unix servers from Sun, Fujitsu and IBM"

Source: http://dev2dev.bea.com/blog/hstahl/archive/2006/05/new_jrockitbase_1.html

PS: Note the first comment ;)

[BitpowerPM]

:banana: :banana:

Maybe a Dual Dual-Core Xeon Woodcrest 5120 1,87 GHz FSB1066 pumped to FSB1333 and 2,33 GHz (or more) and Quad-Channel FB-DIMM DDR2-667 is the better choice than Conroe E6700 @ 3,x GHz. :D :D

[Kjaks]

:banana: :banana:

Maybe a Dual Dual-Core Xeon Woodcrest 5120 1,87 GHz FSB1066 pumped to FSB1333 and 2,33 GHz (or more) and Quad-Channel FB-DIMM DDR2-667 is the better choice than Conroe E6700 @ 3,x GHz. :D :D
No, I don't think so. The Core2 arcitechture doesn't gain much performance from a higher bus than 1066, that's according to a Conroe VS Merom comparision made by Coolaler where both were running at 3GHz, but the Conroe with 1500MHz FSB and the Merom with 1000MHz FSB. All scores were so to say the same. A higher multipler, lower bus speed and higher core speed seems to be a better choice than a low multipler and high bus speed, but lower core speed. But the FB-DIMM is interesting tho.

[TarTheDark]

Not gaining much performance from a higher FSB => Core 2 isn't really bandwith starved, right? :D Not like NetBurst, anyway.

Talking about NetBurst here's an interesting quote from an article on the Inq:

"One of my contacts raised a possibility that, actually, the NetBurst clock speed tricks brought over into the Core 2 Duo design could, even with simple modular water cooling like Corsair Nautilus, enable consistent clock speeds pretty similar to this 4.26 GHz Presler XE in my house, but at the Conroe per-clock efficiency - right this year!"

Source: http://www.theinq.com/?article=32062