Original URL: http://www.channelregister.co.uk/2008/02/06/intel_tukwila_2bn/
What with being the keeper of Moore's Law and all that, Intel found its rightful place as the first revealer of a 2bn transistor chip this week at the sausage fest that is the International Solid State Circuits Conference in San Francisco.
The chip goes by the name Tukwila, and it's a four-core version of Itanium due out in the second half of this year. Lucky for Intel and its customers, Tukwila will replace the Montvale chip released with no fanfare (http://www.theregister.co.uk/2007/11/08/hp_montvale_itanium/) late last year. The fresh chip should provide more than twice the horsepower of Montvale on a variety of software jobs.
But is twice the performance really that impressive?
Well, Tukwila has twice the number of cores as Montvale and a ton of transistors or more specifically 2.05bn. Tukwila beats Montvale (24MB) and just about every other chip on the planet with a 30MB cache as well.
According to the ISSCC slides, Intel will ship Tukwila in four different flavors, ranging from 1.2GHz on up to 2.0GHz and 130 watts to 170 watts. Montvale clocks in at 1.66GHz.
And, of course, Tukwila will be the first Itanic to use Intel's QuickPath interconnect - providing 19.2GB/s of peak bandwidth - and integrated memory controllers. As Intel tells it, QuickPath gives Tukwila 6x the peak memory bandwidth and 9x the compute bandwidth when compared to Montvale.
Um, so, Intel's bragging about twice the performance on a couple of benchmarks even though Tukwila has about twice everything of its predecessor and consumes 25 per cent more power, while being manufactured at 65nm instead of 90nm. Are you more moved or dismayed? ®
Intel to tell all about roaring 96GB/s QuickPath interconnect (30 January 2008)
http://www.channelregister.co.uk/2008/01/30/intel_itanium_quickpath/
HP counters Intel's 'Montvale' omerta with love (8 November 2007)
http://www.channelregister.co.uk/2007/11/08/hp_montvale_itanium/
Intel saddles HP with new Itanium (31 October 2007)
http://www.channelregister.co.uk/2007/10/31/intel_montvale_itanium/
Pesky hack divulges Intel's 'Project Copy Hypertransport' (28 August 2007)
http://www.channelregister.co.uk/2007/08/28/intel_csi_kanter/
HP puts customers to another Itanium test with aging gear (2 August 2007)
http://www.channelregister.co.uk/2007/08/02/tukwila_roadmap_hp/
Intel's 'Montvale' half the chip it used to be (2 August 2007)
http://www.channelregister.co.uk/2007/08/02/intel_montvale_itanium/
Intel's many-core version of Itanium to skip 45nm (14 June 2007)
http://www.channelregister.co.uk/2007/06/14/intel_itanium_kittson/
Intel gives up on super-charged 'Gesher' (17 April 2007)
http://www.theregister.co.uk/2007/04/17/intel_gesher_sandybridge/
Intel joins the herd with slimmer, four-core Itanium (6 May 2006)
http://www.theregister.co.uk/2006/05/06/tukwila_four_cache/
Intel's CSI to outperform AMD's Hypertransport (12 December 2005)
http://www.channelregister.co.uk/2005/12/12/intel_csi_low/
Intel delays and slows dual-core Itanium (24 October 2005)
http://www.channelregister.co.uk/2005/10/24/intel_montecito_delay/
Intel to add memory controllers to future Xeons, Itanics (16 June 2005)
http://www.theregister.co.uk/2005/06/16/intel_integrates_mem_control/
Intel confirms Itanium has a future (2 March 2005)
http://www.theregister.co.uk/2005/03/02/intel_richford_poulson/
Intel puts Itanium saviour on ice (24 February 2005)
http://www.theregister.co.uk/2005/02/24/intel_nicks_tukwila/
Intel admits Itanium pains, plots server future (8 September 2004)
http://www.theregister.co.uk/2004/09/08/intel_whitefield_arrives/
Who sank Itanic? (17 February 2004)
http://www.theregister.co.uk/2004/02/17/who_sank_itanic/
Multicore Itanic: Call me Tukwila (18 December 2003)
http://www.theregister.co.uk/2003/12/18/multicore_itanic_call_me_tukwila/
© Copyright 2008