Info: Amd May Intro 5mb Z-ram Cache

[athlonium 0]

EE Times has details on AMD getting a license for Innovative Silicon's SOI memory. Are we looking at embedded Z-RAM on next generation CPUs from AMD?

It could also provide a means to help AMD put distance between itself and PC processor rival Intel Corp. which, so far, has said it does not see a need to go to SOI technology. AMD's licensing of Z-RAM could also have implications for IBM and Chartered Semiconductor, with which AMD is aligned.

“The dramatic increase in density offered by ISi’s Z-RAM embedded memory can enable much larger on-chip microprocessor cache memories resulting in improved performance and reduced I/O power consumption,” said Craig Sander, corporate vice president of technology development at AMD.

Z-RAM promises 5 times the density of SRAM so could we see dual core chips with 5MB of L2 per core? The story also alludes to this tech showing up in AMD CPUs as soon as two years from now. In two years Fab 36 will be at full capacity, and 65nm will be well in hand. I for one wouldn't mind an ?FX60 with 10MB of L2 cache, or perhaps by then a quad core FX100 with 40MB of L2? On the Innovative Silicon site you can find white papers and an independent analysis for their technology. Apparently it is good enough to get them financial backing from Austin Ventures. We will see what else we can discover about Z-RAM and report what we find out.

Posted by: Chris_Tom on Jan 19, 2006 - 07:54 AM - AMDZone

similar articles:

http://www.eetimes.com/news/latest/showArt...cleID=177101749

http://www.theinquirer.net/?article=29106

[smile4me 1]

klu L2 5mb cache per core,best giler tp harga kompem berganda gak

[athlonium 0]

ni SOI memory.... hmm rasanya xde la mahal sangat...tapi yang penting performance pasti improve gilerr

“The dramatic increase in density offered by ISi’s Z-RAM embedded memory can enable much larger on-chip microprocessor cache memories resulting in improved performance and reduced I/O power consumption"

[Balthazor 1]

ni SOI memory.... hmm rasanya xde la mahal sangat...tapi yang penting performance pasti improve gilerr

←

aaa....aku x paham sgt..ko maksudkan microarthicture ke?.....cam northbridge di integratedkan ke dlm CPU K8 AMD cam tu ker?

[athlonium 0]

keh3...aku pon x paham sebab tech baru....5mb cache tanpa pembesaran core...core ttp kecil.... so cost ttp kurang.....

Advanced Micro Devices, the world’s second largest maker of microprocessors, has singed an agreement with Innovative Silicon Inc., a startup that has developed zero-capacitor memory (Z-RAM) technology that allows to increase processor caches by up to five times without increasing the size of the die.

The Z-RAM technologi is capacitor-less, single transistor DRAM harnessing the floating body effect of Silicon on Insulator (SOI) devices. This technology is capable of achieving twice the memory density of existing embedded DRAM technology and five times that of SRAM, which is used for processor caches, yet requires no special materials or extra mask/process steps. The Z-RAM technology requires silicon-on-insulator process technology and thus will suit for AMD’s central processing units (CPUs).

Z-RAM ni hanya sesuai utk processor yang guna SOI process technology...so kene betul lah ngan AMD...

It could also provide a means to help AMD put distance between itself and PC processor rival Intel Corp. which, so far, has said it does not see a need to go to SOI technology.

intel sampai skrang x nak adopt SOI...

Edited January 20, 2006 by athlonium

[athlonium 0]

more info ttg Z-RAM

http://www.theregister.co.uk/2006/01/20/amd_licenses_z-ram/

[ahmades 2]

Mak aih 5 MB cache on-die dalam proc AMD!!!? Pergh... sangat lawa dan cantekss punya. Nih yang menyebabkan kau nak bertukar pada AMD nih. Bila agaknya Intel nak bagi 5MB gaks?

[vampire 0]

makin lame makin terok intel........

[athlonium 0]

tp aku still blur ttg tech ni .... tapi dia embed masa process SOI .... tapi tanpa capacitor...blur2...dijangka akan launch utk pengguna utk 45nm process...amd tgh test utk 90nm and 65nm dulu....so lama lagi.. aku rasa persediaan utk apa2 yang intel launch.... kira senjata simpanan la nih

[TonikCapGajah2013 175]

sekarang nie intel dah tenat dah...he...he.

[athlonium 0]

citer pasal tenat...ade jumpe satu article:

TG Daily interviews AMD: "Intel's new architecture is too late"

http://www.tgdaily.com/2006/01/20/tgdaily_...ws_amd_q1_2006/

[Balthazor 1]

citer pasal tenat...ade jumpe satu article:

TG Daily interviews AMD: "Intel's new architecture is too late"

http://www.tgdaily.com/2006/01/20/tgdaily_...ws_amd_q1_2006/

←

bukan intel nak wat 100 core in single chip pada thn 2010 nanti ke...........pastu siap kata nak integratedkan memory (RAM) yg kite pakai skrg ni kat CPU dia.............link dia he he he aku tgh carik..........

[athlonium 0]

x logik la... bapak 100 core bape besar packaging

http://www.theregister.co.uk/2005/03/04/intel_100_core/

ni aku rasa saje nak tunjuk power je...xde bukti pon... hahah dulu nak naik ramp up pentium 4 sampai 10Ghz pon tak tercapai... siap nak bubarkan architecture netburst lagi.... huhu this year intel nak launch new architecture pon tak dengar citer... Pentium is dead!... ahhahaa..... intel tak nak guna dah nama tu...

processor terakhir netburst dah launch.... tapi still x leh beat athlon X2 ... hhehehe

dlm apache benchmark pon x leh beat 3800+ X2...

Intel's Last Hope Reviewed : Pentium D 940

Netburst’s Last Hurrah : Intel’s Pentium-D 900 Series

Edited January 23, 2006 by athlonium

[athlonium 0]

Curious to know if all that cache helps the Pentium-D in a server environment? Be curious no longer. It doesn’t.

huhu games, ngan server environment mmg sesuai guna Athlon X2....

Edited January 23, 2006 by athlonium

[HanevE 1]

tp aku still blur ttg tech ni .... tapi dia embed masa process SOI .... tapi tanpa capacitor...blur2...dijangka akan launch utk pengguna utk 45nm process...amd tgh test utk 90nm and 65nm dulu....so lama lagi.. aku rasa persediaan utk apa2 yang intel launch.... kira senjata simpanan la nih

←

weh ape jenis Cache tanpa capasitor.. camne nak cycle power.. data tak loss ke nnt..? Z-RAM ..? stand 4 (zero-capasitor dram)....

SOI ?? silicon-on-insulator...

komen aku..

tech ni bagus sebab....

-dapat recude core cell memory tu. so density akan bertambah.

-cost per bit akan menurun sebab no capasitor..

tapi aku masih tak faham tech ni..

tunggu je la IS lunch kan Z-RAM ni.

[athlonium 0]

yup... AMD mmg pakai SOI process.... IBM pon pakai utk Core cell... processor yang pakai SOI ni kurang leakage power...lebih sejuk ...

[FryShadow 90]

wallane...makin power AMD nih..tp ako tak sempat gak nak tukar proc nie..tp tetap penyokong AMD walaupun gna intel ahaks...benci intel ako nie!!

pasal la da naik rege AMD kat market m'sia nie....

[athlonium 0]

permintaan tinggi skrang amd nih....supply kurang....tu naik harga tu

[polyfuze_4336 0]

hmm 5MB L2 cache..very interesting..embedded kat core..so tak guna banyak space..possibly we could see increase in L2 cache,and it would be interesting too to see the use or should i say the non use of capacitor to hold the charge..

but in 2006 or 2007 Intel will release Montvale with 24MB of L2 cahce and multithreading..even now Intel are using montecito processor for high end server with 24MB of cache and EPIC architecture not the usual RISC or CISC

ahhh bestnyer tgk competition antara 2 rival..sbb nanti user gak yang dapat benefit nyer

[athlonium 0]

ala biasa la tu.... 1mb amd masih bleh beat 4mb cache intel nya proc.... hehhehehe...apa-apa pon montvale tak bersaing utk x86 segment... Itanium architecture pon masih blom baik penerimaannya....

Montvale is the code-name of a future release of Intel's Itanium Processor Family (IPF), which implements the IA-64 instruction set architecture. It is not yet known what the official marketing name for the product will be.

Current estimates suggest that Montvale will be released approximately one year after Montecito, or approximately the fourth quarter of 2006. Considering Montecito's numerous delays, it may slip further as well.

In contrast to earlier speculations, Montvale will not shrink the die to 65nm, but it will cover a Montecito on-steroids. Earlier data suggested that Montvale's clock speed would have likely hit 2.5-2.6 GHz, sitting on a 400 MHz FSB. More recently, people speculate that the processor may not reach default clock speeds in excess of 2.0 GHz.

It is also very likely that Montvale, as Montecito did before it, will bring an update of the compiler technology, with significant improvement in performance characteristics to IPF.

Edited January 31, 2006 by athlonium

[polyfuze_4336 0]

ala biasa la tu.... 1mb amd masih bleh beat 4mb cache intel nya proc.... hehhehehe...apa-apa pon montvale tak bersaing utk x86 segment... Itanium architecture pon masih blom baik penerimaannya....

←

did you read that somewhere?or is that your personal opinion?1mb beat 4mb..could you provide the link on where did you actually read that?

thanks

[athlonium 0]

carik la sendiri...banyak... kat google...

intel yang melambak2 cache pon masih perform kurang dari amd....

1mb x 2 athlon x2 easily beat Pentium D 2mb x 2

[polyfuze_4336 0]

conclusion for benchmark test between opteron and xeon by tomshardware

Conclusion: Linux Servers Smile, And Workstations Worry

In the extensive benchmark tests under Linux Enterprise Server 8 (32 bit as well as 64 bit), the AMD Opteron made a good impression. Especially in the server disciplines, the benchmarks (MySQL, Whetstone, ARC 2D, NPB, etc.) show quite clearly that the Dual Opteron puts the Dual Xeon in its place.

It looks a bit different in the workstation tests, where the Dual Xeon manages to overtake the Opteron team. However, these results are only included because the Dual Opteron will also be available as a workstation option. They give us an interesting picture of the performance of the Opterons compared to desktop CPUs like Athlon XP and Intel P4.

Opteron made a good start out of the blocks - even though AMD left it very late to get its systems into reviewers' hands. With a good showing in the 64 bit arena, albeit with limited benchmarks, AMD has come some way towards regaining the confidence of its supporters. Now, the company can hammer home the key selling point of this strategic move. Compatibility for existing x86 software is going to be the main mantra for AMD. This will save developers and companies money, especially in the business sector. Migrating to 64 bit can therefore take place gradually and only as required. Whether there is enough demand for migration is another matter entirely.

Regarding the architecture of the Opteron, AMD has integrated the 64 bit extensions into the existing x86 concept smoothly and efficiently. There are few problems in moving to 64 bit under Linux: the operating system has been 64 bit capable for years (starting from kernel 2.0), and only the kernel has to be replaced anew. In Windows 64 bit is another issue altogether. Still, the incorrigible complainers and pedants won't have to worry: apart from a very few exceptions, 32 bit software will continue to run without any problems.

The Hammer architecture, compared to the previous Athlon, does not offer any huge changes in the core design or the command processing. More emphasis was placed on optimizing it with regard to higher clock frequencies. A 12-level pipeline is supposed to work together with SOI technology to create potential for higher frequencies. The model numbering system for server CPUs is worthy of some criticism: the number 244 (the current top model with 1.8 GHz) is truly confusing. And it won't be of any extra help to see the notes in AMD's documents that 44 is faster than 42 but slower than 46. How about interpreting the "44" as 4400+ in the rating?

In the end, this number has to have some sort of concrete relationship to the actual performance of the CPU. A maximum frequency of 2.4 GHz (model 52) should be available with SledgeHammer in the mid-term. In any case, numbering has to be made more transparent, so that the clientele can have a point of comparison. However, it must be said, the server customer is not interested in performance by itself, but in stability, software development, support, and availability.

the conclusion is quite predictable, AMD excel in some areas while Intel in another..there is no clear cut winner in the benchmark..in memory benchmark Intel won the test,while AMD won the FPU test..dari segi performance masih fair..50-50..nothing extraordianry performance leap both from Intel and AMD

nanti aku baca lagi pasai nie..and latest AMD vs Itanium2

[athlonium 0]

tp ko tgk la apache bench kat atas tu...

hmm...Itanium lain architecture bkn x86....

itanium ni masih peringkat projek kerjasama intel ngan hp.... software pon OpenVMS rasanya... tp ade org run windows over OpenVMS....

Edited January 31, 2006 by athlonium

[athlonium 0]

http://www.extremetech.com/article2/0,1697,10350,00.asp

64bit : AMD Hammer vs Intel Itanium