I've been slowly building this post, collecting all of the available info so far (pricing, energy consumption, benchmarks, scientific background, release dates, etc...). If anyone has any credible benchmarks they would like to see added to this list, please either PM me a link or post it in this thread. Credible benchmarks where something beats Conroe are most welcome, as this post should represent all available information. I just haven't seen any benchmarks where it loses yet.

Yes this is pretty much a fanboy post if you want to call it that, but I want to make it very clear that I do not care what company makes it. I am a fan of this particular microarchitectural design and implementation. I don't care what it says on the box it comes in: Intel, AMD, Via, TI, Samsung or otherwise...

That said, here is what I have so far:

Post Release update August 4th 2006:

I wanted to update this because this is just too cool. Coolaler has pushed his Conroe to 5.5GHz and got a 9.000 second SuperPi_1M run. He is a truly talented overclocker with a truly amazing chip. This makes the Conroe WR more than 100% faster than any other consumer microarchitectural design WR in this benchmark. Check it:






Official Core 2 Reviews And Info

Well the NDA is up and all of these sites have independently verified what we have known for the past couple of months. Core 2 is the best chip you can buy, period. Performs the best, uses less energy, produces less heat, has better performand per watt as well as performance per dollar. Here are a sample of reviews (in alphabetical order)
AnandTech- http://www.anandtech.com/cpuchipse [...] spx?i=2795
Bit Tech - http://www.bit-tech.net/hardware/2 [...] ors/1.html
ByteSector - http://www.bytesector.com/data/bs-article.asp?id=661
Chile Hardware - http://www.chilehardware.com/modul [...] =200607131
Club IC - http://www.clubic.com/article-3635 [...] 2-duo.html
Computer Base - http://www.computerbase.de/artikel [...] eme_x6800/
Digit Life - http://www.digit-life.com/articles [...] e6600.html
Extreme Tech - http://www.extremetech.com/article [...] 036,00.asp
Firing Squad - http://www.firingsquad.com/hardwar [...] rformance/
GD Hardware - http://www.gdhardware.com/hardware [...] 00/001.htm
GotFrag - http://hardware.gotfrag.com/portal/story/33492/
Guru3D - http://forums.guru3d.com/showthread.php?t=185555
HardOCP - http://enthusiast.hardocp.com/arti [...] VzaWFzdA==
Hardware Secrets - http://www.hardwaresecrets.com/article/348
HardwareZone - http://www.hardwarezone.com.sg/art [...] =2&id=1980
Hardware.fr - http://www.hardware.fr/articles/63 [...] i975x.html
Hexus - http://www.hexus.net/content/item.php?item=6184
Hot Hardware - http://www.hothardware.com/viewart [...] =845&cid=1
Legion Hardware - http://www.legionhardware.com/document.php?id=569
Legit Reviews - http://www.legitreviews.com/article/362/1/
MadBox PC - http://www.madboxpc.com/contenido.php?id=2394
Mad Shrimps - http://www.madshrimps.be/?action=g [...] rticID=470
Maximo PC - http://www.maximopc.org/articulos/ [...] rte_1.html
NeoSeeker - http://www.neoseeker.com/Articles/ [...] duo_e6700/
OCAU - http://www.overclockers.com.au/article.php?id=489587
OC Workbench - http://www.ocworkbench.com/2006/intel/core2duo/g1.htm
PC Perspective - http://www.pcper.com/article.php?aid=272
Phoronix - http://www.phoronix.com/scan.php?p [...] =512&num=1
Planet X64 - http://www.planetx64.com/index.php [...] &Itemid=14
Sharky Extreme - http://www.sharkyextreme.com/hardw [...] hp/3620036
Sim HQ - http://www.simhq.com/_technology2/technology_090a.html
Tech Report - http://techreport.com/reviews/2006 [...] dex.x?pg=1
Tom's Hardware - http://www.tomshardware.com/2006/0 [...] athlon_64/
Trusted Reviews - http://www.trustedreviews.com/article.aspx?art=3161
TweakTown - http://www.tweaktown.com/articles/923/
Xbit Labs - http://www.xbitlabs.com/articles/c [...] e6300.html


Benchmarks / Chronicle of Records: NOTE: many of these older benchmarks are all with A0 stepping Core 2's, meaning they are only the first run in the fabs. Production products (rumored to be B0 or later) will meet or exceed these results, clock for clock. After mid-May we are starting to see B0 stepping Core 2 Duos and Extremes. Here is a collection of benchmarks available so far...

June 29 2006: Tons of new Core 2 Duo info is now available and it all agrees with what we've known for the past couple months. Here is a complete gaming and media encoding/compression benchmarking suite for a few of the Intel Core 2 Duo Products, with the AMD FX-60 and FX-62 scores for comparison, and a 3.66GHz Core 2 Extreme Air cooled overclock thrown in for fun.

The data speaks for itself. Core 2 Dominates.



and by the way... history was made today.



The Core 2 Extreme X6800 has been overclocked to 5.2GHz by Coolaler and set a SUB 10 SECOND SuperPi_1M run! The record now stands at 9.828 seconds

June 16 2006: Well they said it could be done and now we have proof. Conroe can hit 4GHz on Air cooling alone! Denny has gotten his hands on a new Stepping-5 (likely retail sample) of an E6700 Core 2 Duo and paired it with his Tower 120 Air Cooler.

Link: http://www.xtremesystems.org/forum [...] p?t=103381

13 Second SuperPi_1M run on air cooling...

June 15 2006: FreeCableGuy has a new Stepping-5 Core 2 Duo E6700. Which is an even newer fab run than anything we have seen thus far. And it has improved again both in performance per clock and stability as less voltage is needed at high overclocks. As has been explained before, the launching products are acutally better than most of the engineering samples we have seen thus far, as Intel tweaks their manufacturing processes.

He is curently limited by his motherboard (426MHz FSB) and still with that he has hit 4.3GHz with this chip and set some new world records.

Aquamark3: 202,954
3dMark05: 23,664
SuperPi: 12.281 @ 4.26GHz @ 1.457 Vcore!

Now this isn't a world record SuperPi time (that was set with a 5GHz Core 2) but the key is that it achieved this amazing score with a 1.475 Vcore. That is truly amazing.

Link: http://www.xtremesystems.org/forum [...] p?t=103349

June 11 2006:
Some new Core 2 Duo E6700 and X6800 benchmarks:
http://www.hothardware.com/viewart [...] =832&cid=1

Here we see Conroe besting an overclocked (2.8GHz) FX-60 system in benchmarks such as FutureMark PCMark05 and 3DMark06, as well as real life workloads such as Quake 4 single threaded and multithreaded, low and high resolutions, FEAR, Farcry, LAME Audio encoding, DIVX encoding, and Photoshop filtering. The 2.66GHz Core 2 consistantly wins by 15-30% and the Core 2 Extreme wins by larger margins, as expected.

June 10 2006:
Tarosa has set new World records in SuperPI 2M-32M:

M/B: Intel D975XBX Rev.303 mods BIOS 1181 2006.05.26
CPU: Conroe E6700 ES
Memory: Patriot PC2-8000 XBLK CL4 1GB ×2
HDD: Seagate ST318453LW 18GB ULTRA320/Adaptec 39320D 32Bit PCI
VGA: KUROUTO SHIKOU　FX5200 PCI 128MB
PSU: Zippy-600W the Gaming Edition 3.3V Rail mod →3.45V
OS: Windows Server 2003
Vcore:1.731V, Vdimm:2.91Ｖ(Actual), Vnb:1.725V(BIOS), Vfsb:1.395V(BIOS)

2M: 26.921s http://www.geocities.jp/gbxfs134/26_921.jpg
4M: 1m03.562s http://www.geocities.jp/gbxfs134/103562.jpg
8M: 2m28.329s http://www.geocities.jp/gbxfs134/228329.jpg
16M: 5m22.797s http://www.geocities.jp/gbxfs134/5_22_797.jpg
32M: 11m51.797s http://www.geocities.jp/gbxfs134/11_51_797.jpg

for reference, here are the AMD world records:

2M: 48.750s http://green.ap.teacup.com/fredyama/img/1137895193.gif
4M: 1m 47.781s http://premium1.uploadit.org/fredyam...3893.9MHzR.gif
8M: 3m 55.156s http://premium1.uploadit.org/fredyam...-3893.9MHz.gif
16M: 8m 47.218s http://premium1.uploadit.org/fredyam...3859.9MHzR.gif
32M: 18m 45.156s http://premium1.uploadit.org/fredyam...-3893.8MHz.gif


June 9, 2006:
Core 2 Duo E6700 based system has set new world records for 3DMark01SE (74353), 3DMark05 (23851), and Aquamark3 (201,602).

Links:
http://www.oohashi.jp/c-board/file [...] 308.87.png
http://www.oohashi.jp/c-board/file [...] 308.87.png
http://www.oohashi.jp/c-board/file [...] 166.67.png


June 6, 2006: The independent reviews are starting to hit the web. Here Anandtech.com got their hands on a Core 2 Extreme and put it up against their FX-62 in their hotel room. To keep the tests as fair as possible, they used the same RAM, graphics card, harddrive, etc... They also ran them both at their stock shipping speeds. If any of the evidence below in this thread is valid, when overclocked, the dominance of Core 2 will only extend further.

They ran their own benchmarks and were not guided by Intel in any way.

Link: http://anandtech.com/cpuchipsets/s [...] i=2771&p=1

And here is a quick analysis of their data with additional data for the older Intel flagship processors thrown in for comparison. Needless to say, Core 2 Extreme wins across the board:

Edit: updated the image slightly to fix a typo in the FX-62 price, and slightly off TDP data for the Pentiums. Should be correct now.

June 4, 2006: Here we have a new comparison of Core 2 Duo E670 versus Intel's current Flagship offering the Pentium XE 965: http://www.bit-tech.net/hardware/2 [...] iew/1.html

New WR time for SuperPi_1M: 10.140 seconds @ 5013.1Mhz this weekend:


June 1, 2006: Here are some new Conroe XE 3.2GHz, E6800, P-D EE 965 and FX-62 benchmarks and comparisons:
http://www.extremetech.com/article [...] 191,00.asp

May 30, 2006: Core 2 Extreme has officially broken 5GHz and a new SuperPi_1M record has been set (10.281sec):

Link to news thread: http://www.xtremesystems.org/forum [...] p?t=101421

May 29, 2006: Complete analysis of Conroe.Merom SuperPi performance:


May 28, 2006 Records: TAM is about to break 5GHZ!!!!
SuperPi_1M: 10.312 seconds on a Conroe E6800 @ 4.99GHz
Link: http://www.xtremesystems.org/forum [...] tcount=122

May 26, 2006 Records:
SuperPi_1M: 10.750 seconds <-- new world record
TAM got his hands on an E6800 (2.93GHz stock) Conroe and has overclocked it to.... 4.74GHz. And this is just the beginning.... this was in an INTEL motherboard at 1.324 Vcore, lol.
Link: http://u-san.net/c-board/file/X6800-4739_10s750.gif

May 25, 2006 Records:
The B0 Conroes are Overclocking Monsters! Everything they throw at it, this chip takes it in stride. No cold bugs to be found! Coolaler is runing a new clockgen, fsb overclock, etc.. and is reaching 4.6GHz overclocks!
SuperPi_1M: 11.359 seconds <-- new world record
SuperPi_2M: 28.812 seconds <-- new world record
SuperPi_4M: 1:08.625 <-- new world record
SuperPi_8M: 2:35.375 <-- new world record
SuperPi_16M: 5:45.703 <-- new world record
SuperPi_32M: 12:29.312 <-- new world record
Link: http://www.xtremesystems.org/forum [...] tcount=127

May 24, 2006 Records:
Coolaler has gotten a newer revision of the MSI Conroe motherbord that actually has support for Vcore changes, and now he can overclock even higher using his new 2.66GHz B0 Stepping Core 2 Duo.
SuperPi_1M: 11.922 seconds <-- new world record
SuperPi_2M: 29.968 seconds <-- new world record
SuperPi_4M: 1:11.031 <-- new world record
SuperPi_8M: 2:39.906 <-- new world record
SuperPi_16M: 5:52.344 <-- new world record
SuperPi_32M: 12:44.500 <-- new world record
Link: http://www.xtremesystems.org/forum [...] stcount=33

Here VictorWang makes another appearance with his Conroe and amazing overclocking grapohics cards...
3DMark05: 21442 <-- new world record
3DMark01se: 64581 <-- new world record
http://www.xtremesystems.org/forum [...] p?t=100456

Hexus has gotten their hands on a 2.6GHz Conroe system and did some benchmarking against a real AMD-supplied 2.8GHz FX-62. We should note that the Conroe test system used had a drastically less capable memory configuration (in number, clock and total capacity of DDR2 DIMM), as compared to that of the FX-62 test system. The comparative results for Conroe may well have been even better in certain of the tests, had the memory configurations been equivalent. I've quoted the real-world benchmarks below:
Benchmark/System: % Faster/slower than AMD Athlon 64 FX-62
Realstorm Raytracing 2004: 19.09 faster
DivX encode - multithreaded: 31.20 faster
WAV conversion multithreaded: 32.46 faster
CINEBENCH multi-CPU render: 15.77 faster
KribiBench v1.1 - Jetshadow model: 51.32 faster
Far Cry - 1024x768 - speed: 39.47 more
Quake 4 - 1600x1200 - 4x AA 16x AF: 3.67 more
Splinter Cell: CT - 1600x1200 - 4x AA 8x AF: 0.88 more
Link to article: http://www.hexus.net/content/item. [...] 692&page=1


SuperPi_1M on a 2.66GHz B0 stepping Conroe @ 4.1GHz: 12.484 seconds <--
http://coolaler.kj.idv.tw/WR/4098_1.gif
Coolaler has figured out a pin mod for his motherboard so he can start to up the voltage, showing some the potential of Conroe when it is on a motherboard where you can control the voltage. I have a feeling this (4.1GHz) is just the begining...

SuperPi_1M on a 2.66GHz B0 stepping Conroe @ 4.0GHz: 12.609 seconds <-- http://coolaler.kj.idv.tw/WR/4000_1M1.gif
Coolaler has pushed his Conroe to a stable 4.0GHz overclock at stock voltage. This chip is a beast. Imagine what will be possible with a multiplier unlocked XE on a fully conroe supporting motherboard...

SuperPi_1M (and many other) records broken by 2.66GHz B0 stepping Conroe @ 3.8-4.0GHz: 12.984 seconds SuperPi_1M <-- http://www.xtremesystems.org/forum [...] stcount=68
As promised, a 2.66GHz B0 stepping Conroe has been made available, and it is overclocking even higher. Coolaler took every single SuperPi time with his new chip. All of the records will fall, as it is clear that this stepping is much better clock for clock, and can handle mush higher stable overclocks. Such as the 3.8GHz overclock at stock 1.267 Vcore above...

SuperPi_1M (and other) records broken by 2.4GHz A1 stepping Conroe @ 3.6GHz: 13.922 seconds SuperPi_1M <-- http://www.xtremesystems.org/forum [...] hp?t=99892
So pretty much so far all of the benchmarks and records I've been collecting here are based on 2.13GHz Conroe and Merom engineering samples. FUGGER has in his possession, a 2.4GHz stock clocked Conroe, which he is now demonstrating another jump in performance. This chip, as expected, overclocks in a stable manner even higher than the 2.13GHz base chips. Follow the link to see his score for Cinebench 32-bit, CrystalMark, PiFast, 3dMark01, 3dMark06, 3dMark05, Aquamark3, and SuperPi_1M. Rumor has it that samples of some of even higher clocked stock versions of these chips will start becoming available soon. (Possible 3.33GHz Core 2 XE)

SuperPi_1M Record broken again by Merom @ 3.5GHz: 14.500 <-- http://coolaler.kj.idv.tw/merom/QLZT/3552.gif

And again: Merom @ 3.6GHz: 14.313 seconds <-- http://coolaler.kj.idv.tw/merom/QLZT/3591.gif

And again: Merom @ 3.66GHz: 14.031 seconds <-- http://coolaler.kj.idv.tw/merom/QLZT/3655.gif

New World Records (re)set by 3.2-3.4GHz Core 2 Duo (Merom):
SuperPi Record broken again: 14.688 <-- http://coolaler.kj.idv.tw/merom/QLZT/merompi.JPG
Aquamark3 Record broken again: 180,329 <-- http://222.151.144.54/c-board/file [...] S_3434.png
3DMark05 World Record: 20672 <-- http://222.151.144.54/c-board/file [...] 7400ES.png

New World Records set by 3.2-3.4GHz Core 2 Duo (Merom): http://www.xtremesystems.org/forum [...] stcount=51
SuperPi_1M: 14.734 seconds <-- World Record http://222.151.144.54/c-board/file [...] 3440.6.png
Hexus Pi_Fast: 27.66 seconds <-- World Record http://222.151.144.54/c-board/file [...] 3408.4.png
Aquamark3: 177,543 <-- World Record http://222.151.144.54/c-board/file [...] S_3382.png
3dMark05: 20380 <--- Almost the world Record http://222.151.144.54/c-board/file [...] 7400ES.png

Media Encoding (NEW!): http://www.xtremesystems.org/forum [...] stcount=12
Brand new data, here we see the 2.13Ghz Stock Conroe system encoding a video using TMPGEnc 3.0 Express in 20:54, while it takes an overclocked 2.9GHZ FX-60 21:06 to encode the same exact video. I'll state that again. A 2.13GHz Conroe is faster than a 2.9GHZ FX-60 in media encoding! I'm looking forward to a clock-for-clock comparison.

EDIT: here is the clock for clock comparison: http://www.xtremesystems.org/forum [...] tcount=188
Recap: 2.9GHZ FX-60 takes 21:06. 2.9GHz Conroe takes 15:22 for the same encoding task.

F.E.A.R: http://www.bit-tech.net/news/2006/ [...] mark_fear/
At this site the guys at Bit-Tech loaded their own personal FEAR benchmark from a USB thumbdrive, so that they could compare the results from their own tweaked FX-60 system outside of the control of Intel. Compared to their FX-60, with very tight memory timing that they spent days tweaking, the Conroe system get's 36% more FPS in this CPU bottlenecked gaming benchmark.

Half Life 2: Lost Cost: http://www.xtremesystems.org/forum [...] hp?t=98091
Here FCG has set a world record by a HUGE margin, there are no systems other than Conroe based systems that have been able to post anywhere even remotely close to 165 average fps in HL2:LC no matter what crazy overclocked graphics system they have.

Media Encoding: http://www.anandtech.com/tradeshow [...] i=2713&p=4
Here a 2.6GHz Conroe is shown besting an overclocked FX-60 (@ 2.8GHz) by anywhere from 12 to 30% in various media encoding tasks.

Quake 4: http://www.hothardware.com//viewar [...] icleid=794
Here Conroe bests the FX-60 by 23-28% in Quake 4 performance.
http://www.anandtech.com/tradeshow [...] i=2716&p=3
Here Anandtech updates the issues that they had with their first run of Quake 4, running their own in-house time demo. Conroe shows a 24-33% advantage over the overclocked FX-60.

3DMark05: http://members.cox.net/kjboughton/20200.JPG
Here FCG uses a "simulated" Conroe XE (2.13GHz ES overclocked to 3.33GHz) to obtain a 20200 score in 3DMark05.

3DMark01: http://members.cox.net/kjboughton/FCG_01details.jpg
I know it is an old benchmark, but it is still in use... don't stress.
Here, FCG uses Conroe to set a new world record in the incredibly CPU intensive "Game 3 Lobby - High Detail" section of 3dMark01.

PCMark05 CPU test: http://www.xtremesystems.org/forum [...] hp?t=97609
In this test FCG compares a 2.6GHz Conroe to a 3.6GHz FX-57 and a 2.6GHz X-2 4400+. The Conroe beats both systems by over 23%.

3DMark05 CPU test: http://www.xtremesystems.org/forum [...] hp?t=97593
In this SINGLE THREADED CPU test, the Conroe beats the severely overclocked FX-57 by 23%.
This link: http://www23.tomshardware.com/cpu. [...] model2=212 provides you tons of reference scores for practically every other processor out there. Conroe completely destroys them all.
Here is some new data: http://coolaler.kj.idv.tw/merom/QLZT/XP64/4.gif

FX-57 @ 3.60GHz: 8146
Conroe @ 2.66GHz: 10038
Merom @ 3.2GHz: 11212

Aquamark3 Benchmark: http://www.xtremesystems.org/forum [...] hp?t=97395
In this demonstration, the air cooled 2.7GHz Conroe system sets a new world record, besting the previous record holder, a C02 single-phase cooled 3.7GHz FX-60 with liquid cooled , overclocked SLI 7900GTXs.

SuperPi 1.5: http://www.xtremesystems.org/forum [...] hp?t=97269
In this demonstration a 2.13GHz Conroe is overclocked to 2.9GHz with regular air cooling and gets 17.5 seconds on the 1M SuperPi run.
http://www.hkepc.com/bbs/news.php?tid=587650
Now this data is iffy at best, as it does not show the checksum for the superPi result, so take it with a grain of salt. However it appears to show a Conroe chip over clocked to 3.1GHz and getting a ~16 second time on SPi1M.

New: Validated 15.937 second SuperPi_1M run @ 3.2GHz: http://www.xtremesystems.org/forum [...] tcount=189
This is 0.04 seconds away from the world record.

New 2: He just tied the world record: http://www.xtremesystems.org/forum [...] tcount=204

New world record! 15s has been broken! 14.984 seconds for SuperPi_1M!!!
http://222.151.144.54/c-board/file [...] 3376.4.png

3DMark06: http://www.xtremesystems.org/forum [...] hp?t=98115
Here the 2.7GHz Conroe based system is shown getting 11411 3DMarks. Granted with very overclocked graphics cards, but it appears that the Conroe system is really unleashing the full power of the Crossfire system (and this is on a poorly supported, pre-release mobo).
http://www.hkepc.com/bbs/news.php?tid=593760
Here is the first sighting of a B0 stepping 2.67GHz STOCK Conroe. Check out the CPU score in 3dMark06! 2282 stock @ 2.6GHz!!! FYI, The overall score is very low becuase it is running the integrated motherboard graphics. I can't wait to see this with a 7900GTX or something!

3DMark06 CPU Test: http://coolaler.kj.idv.tw/merom/QLZT/XP64/5.gif
FX-60 @ 3.2GHz gets 2380 3DMark06 CPU score
Merom @ 3.2GHz gets 2833 3DMark06 CPU score

Cinebench 2003 64-Bit: 1225 <-- http://coolaler.kj.idv.tw/merom/QLZT/27.gif

3DMark03 64-bit: 36000 <-- http://coolaler.kj.idv.tw/merom/QLZT/XP64/2.gif


Pricing: The current road-map pricing is as follows:

3.2GHz "Conroe" gets confirmed by Intel
Intel representatives just contacted DailyTech with the following information:
The Core 2 Extreme processor (Conroe based) will ship at 2.93GHz at Core 2 Duo launch. We will also have a 3.2GHz version by end of the year. And as you know, the Quad Core enthusiast SKU, Kentsfield, is planned for Q1'07. Full Article

Core Extreme Edition Series - 1333MHz FSB, EM64T, EIST, VT, SSE4
Core X6900 - 3.2GHz, 1066MHz FSB, 32+32KB L1, 4MB L2, Dual Core, TDP 75W - $999 - Q4
Core X6800 - 2.93GHz, 1066MHz FSB, 32+32KB L1, 4MB L2, Dual Core, TDP 75W - $999 - Q3, Price reduced by Q4.

Core E6000 Series - 1066MHz FSB, EM64T, EIST, VT, SSE4
Core E6700 - 2.67GHz, 1066MHz FSB, 32+32KB L1, 4MB L2, Dual Core, TDP 65W - $530 - Q3
Core E6600 - 2.40GHz, 1066MHz FSB, 32+32KB L1, 4MB L2, Dual Core, TDP 65W - $316 - Q3
Core E6400 - 2.13GHz, 1066MHz FSB, 32+32KB L1, 2MB L2, Dual Core, TDP 45W - $224 - Q3
Core E6300 - 1.86GHz, 1066MHz FSB, 32+32KB L1, 2MB L2, Dual Core, TDP 45W - $183 - Q3

Core E4000 Series - 800MHz FSB, EM64T, EIST, SSE4
Core E4200 - 1.6GHz, 800MHz FSB, 32+32KB L1, 2MB L2, Dual Core, TDP 40W - $___ - Q4 <-- Allendale


Energy Consumption:

The 1.83, 2.16, 2.40, and 2.67GHz models have a 65W TDP, the 2.93GHz Extreme Edition has an 75W TDP. For reference, Intel's current top of the line Pentium Extreme Edition 965 has a 130W TDP and AMD's FX-62 has a 125W TDP. Even though Conroe seems to completely demolish these chips in performance, it uses nearly half the power.


Articles about the CORE microarchitecture (the basis for Conroe):

http://www.anandtech.com/cpuchipse [...] spx?i=2748 <-- excellent overview of some of the microarchitectural improvements made in CORE.
http://www.anandtech.com/tradeshow [...] spx?i=2711 <-- coverage of the official unveiling of the Core microarchitecture at Spring IDF 2006, San Francisco.
http://arstechnica.com/articles/paedia/cpu/core.ars <-- A "big picture" post on ArsTechnica.
http://www.tomshardware.com/2005/1 [...] uncovered/ <-- the seminal article here at Tom's covering some of the secret plans that lead to Core and some things to come, concerning 45nm.
http://www.realworldtech.com/page. [...] 0906143144 <-- The original in-depth analysis of the Core architecture. This compares it with Yonah and Netburst. Anandtech and ArsTechnica base their architecture reviews on this one.
http://www.intel.com/technology/ar [...] o/demo.htm <-- Intel Marketing flash animation about some of the new features in Core. (link goes to page with Flash).


The "So what makes Core so new?" Section:

Here I will try to present some of the differences between CORE and previous architectures (in particular, the Core Duo, Intel's most current architecture you can buy today).

Conroe features a large 4MB cache, which is a first in consumer x86 products. This cache also has ultra-fine-grained power control, and an intelligent sharing mechanism between the two cores (in fact on some motherboards you will see that you can turn off one of the cores and dedicate the entire 4MB to one core, for more stablity in ultra high overclocks for single threaded apps. I can't provide proof for this statement yet, but you'll just have to wait and see if I'm right). That power control and sharing mechanism is not a walk in the park to design, and while many of the principles are present in Core Duo, this was a complete redesign with more aggressive power saving algorithms and more sensors and much more fine grained control, meaning smaller sections of the cache can be turned off more frequently.

Conroe also features memory disambiguation which allows most Loads to be speculatively executed before Stores. This is a HUGE benefit for some workloads. This is also not present in Core Duo.

Conroe has a 4-issue core, which means that 4 (or in some cases 5, see below) instructions are simultaneously pulled from the instruction cache and fed through the pipeline. Core Duo (and every other x86 product on the market) is 3-issue.

Conroe has macro and micro-ops fusion which allows certain combinations of x86 ops to be combined into one as they are pulled from the instruction cache. Macro-fusion allows the chip to save energy and have higher bandwidth, particularly during popular CMP-JMP operation pairs (if/then or switch statements in most programming languages decode into these instructions). It saves energy because it can literally do the same work with less physical transistor switching as the fused instruction moves through the pipeline. Core Duo only has micro-fusion.

Conroe has a completely redesigned 128-bit wide SSE engine that completes all SSE ops in 1 cycle. Core Duo has nothing close to this, in fact nothing on the market has anything close to this.

Conroe is 64-bit, Core Duo is 32-bit.

The SSE units also weren't the only thing expanded. The FPUs were also widened to 128-bit from 64-bit in Yonah, but since the FPUs share some of their resources with the SSE units the two are interrelated. The ALUs have also been widened to 64-bit from 32-bit which should provide a performance boost in integer operations as well.

Other improvements include increases in L1-L2 transfer bandwidth and the near tripling of bandwidth between the L1 cache and the rest of the core. A lot of that has to do with the widening of the internal data paths from 64-bits in Yonah to 128-bits in Core.

Other minor details include SSE4 support, although Intel may be looking to come up with a new marketing name. Related to the 4-wide issue rate is that all the buffers have been widened to handle the increase. The number of ports has also increased from 6 from 5 in Yonah to better organize the additional execution units and the division of capabilities (Vshuffle, etc.) for each execution unit has also changed somewhat to avoid conflicts. The instruction fetch size has also increased from 128-bit to 160-bit (possibly higher) in order to feed the 4 decoders.

...to be continued.


Conroe/Merom performance comparison:

Now there are preliminary numbers, both chips are clocked to 2.96GHz in all tests, though they have different FSB speeds.

Thread: http://www.xtremesystems.org/forum [...] hp?t=99466

SuperPI_1m
Conroe: 17.125
Merom: 17.344

CPUmark99
Conroe: 418
Merom:413

SiSoft Sandra Dhrystone/Whetstone/Integer/FP/MemInt/MemFloat
Conroe: 27395/18882/163122/88312/7367/5871
Merom: 27515/18720/163693/88016/5478/5481

SiSoft Sanrda Memory Latency Linear/Random/Cache
Conroe: 60ns/14ns/31390
Merom: 78ns/12ns/28816

CineBench 2003 32-bit 1CPU/xCPU
Conroe: 509/941
Merom: 509/940

TMPGenc
Conroe: 15:34
Merom: 16:39

SuperPi_32M
Conroe: 16:36.797
Merom: 18:04.672

3DMark06CPU
Conroe: 2523
Merom: 2574


Chipset/Mobo Support:

975x rev 304+ <-- http://www.xbitlabs.com/news/mainb [...] 75704.html

P965 and G965 (possibly Q965 later) <-- http://www.xbitlabs.com/news/chips [...] 55917.html

Possible limited support on certain i865G boards from ASRock.

Next Gen ATI Board <-- http://www.xtremesystems.org/forum [...] 424&page=3 <-- see FUGGER's post in this thread for detail.


Release Date:

Current estimates puts this chip in stores on July 27th 2006.


Future Roadmap:

http://www.vr-zone.com/?i=3453 <-- shows Kentsfield Quadcore in 1H 2007 and LV and ULV Merom parts for 1H 2007.


Release Branding:

Intel to Unveil Single Brand for New Chips
©Copyright 2006 Associated Press

http://arstechnica.com/news.ars/po [...] -6771.html <-- coverage on Ars Technica

http://www.dailytech.com/article.aspx?newsid=2170 <-- coverage on Daily Tech

http://www.tgdaily.com/2006/05/07/ [...] duo_brand/ <-- coverage on TG Daily


Other links to peruse:

http://www.xbitlabs.com/articles/e [...] 6-3_4.html
http://www.hexus.net/content/item.php?item=4843
http://techreport.com/etc/2006q1/conroe/index.x?pg=1
http://www.digit-life.com/archive.shtml?2006/0308
http://www.xtremesystems.org/forum [...] hp?t=98376 <-- here we see the mobile version of Conroe, called Merom, doing 20% better clock for clock then a Core Duo, where the exact same system was used in both cases, the Merom chip was simply swapped in on the motherboard.
http://www.computerbase.de/news/ha [...] tel_merom/ <-- 2.16GHz Merom against 2.16GHz Yonah in Quake 4, 20% increase clock for clock.
http://www.computerbase.de/news/ha [...] duo_merom/ <-- 2.0GHz Merom against 2.16GHz Yonah in a whole bunch of synthetic tests, merom wins handily everytime. Look really solid.
http://www.pconline.com.cn/diy/fro [...] 90594.html <-- Possibly the first Conroe Extreme Edition seen in the wild. This looks like a very early physical tester chip. There are not even permanent capacitors mounted on the bottom. I doubt the public is supposed to see this.

Just off the top of my head.

Quake 4

http://www.anandtech.com/tradeshow [...] i=2716&p=3

Although people say Intel didn't allow Anand to change anything on the systems, Anand did load his own custom Quake 4 Demo and the results are provided here to compare with Intel's default Demo.

Core Architecture

http://www.realworldtech.com/page. [...] 0906143144

The original in-depth analysis of the Core architecture. This compares it with Yonah and Netburst. Anandtech and ArsTechnica base their architecture reviews on this one.

Thanks. Added them in.

A lot of quality information

This should be stickied! Will stop quite a few BS threads. :!:

Endyen will be uber pissed when he sees that.

Endyen will be uber pissed when he sees that. So will MadModMike. MMM: "There will be NO 3.33GHz Conroe! You fanboys can keep dreaming."

Goes without saying. Though I do doubt it'll be 3.33ghz. We'll wait and see.

I just added a note on the 3.33GHz version pricing info that it may not be available at launch time of the other processors, as it is a relatively credible rumor.

However, it is also worth noting that the overclocking stability of these chips looks very solid. Therefore, once motherboards with proper support are available and some create cooling solutions start being applied, I wouldn't be surprised to see these reaching 3.5-3.8GHz in enthusiast's hands within the first month.

For example, there is a 2.13GHz Engineering Sample that is SPi32M stable at 2.9GHz and 24Celcius with just a prescott air cooled heatsink. Anything more exotic than watercooling may not be neccessary for some serious overclocks.

So will MadModMike. MMM: "There will be NO 3.33GHz Conroe! You fanboys can keep dreaming."[/quote]
Ohhh, yeah....and there will be 4GHz K8, acording to master MMM sources

Bah, I say. They will push it further.