2007 Roadmap

H2 (2nd Half)

Serial ATA III (SATA 3) specification is expected to finalized in H2 2007. The data transfer speed will be doubled to 600MB/s and new connector formats supporting storage arrays or mobile optical swap bays will be added. Support for multiple SATA 3 devices on the same SATA line may also be added.

Q3 (3rd Qtr)

September

AMD Opteron (Code name: Barcelona) was released on September 10th. Barcelona, dubbed the K10 architecture, is a native quad-core CPU with speeds ranging from 1.7GHz to 2.0GHz. All the cores share 2MB of L3-cache. L1 and L2 caches remain the same with 128KB and 512KB respectively. The CPU is available in two different models supporting either 4/8-way (4S) or 2-way (2S) configurations. These new Opterons have substantial architectural differences from the original AMD’s Opteron and Athlon 64. They feature a tweaked cache, memory controller, branch predictors, prefetch logic, power management and additional AMD-V extensions.
ATI HD 2900 PRO was released on September 25th. The HD 2900 PRO comes in two variants, 512MB GDDR3 and 1GB GDDR4. The 512MB variant comes includes 320 stream processors, a 600 MHz core clock, and 800 MHz GDDR3 memory. The 1GB has a 742 MHz core clock. Both cards are available in 9 inch and 12 inch board sizes with the same size fansinks for cooling.
nVidia MCP73 was relased on September 25th. MCP73 is an IGP for Intel processors based on the GeForce 7 series. It will have three variants, the GeForce 7150 and 7100 paired with the nForce 630i and the GeForce 7050 goes with nForce 630i or nForce 610i. These are Vista compatible and support aero. The GeForce 7150 features a core clock speed of over 600 MHz, while the 7100 has a 600 MHz clock, and the 7050 a 500 MHz clock. The 7150 and 7100 support DDR2-800 and the 7050 is limited to DDR2-667. The 7150 and 7100 also support a higher 1333 MHz FSB; the 7050 with a maximum 1066 MHz FSB. All three solutions will have one PCIe x16, two PCIe x1, four SATA 3.0Gbps, two PATA and RAID; however, the nForce 630i will have more USB ports, gigabit ethernet, and more RAID configurations.
AMD Athlon 64 X2 5000+ "Black Edition" was released on September 27th. This "Black Edition" processor is identical to the regular X2 5000+ featuring a 2.6 GHz clock, 1MB L2 cache, and 65W TDP except that it has an unlocked multiplier allowing for frequency overclocks without needing to change the HyperTransport clock. It will sell for $136 in quantities of 1000.

Q4 (4th Qtr)

October

AMD released 6 new 45 watt CPU's on October 8th. The new processors will be marketed towards desktop and home theater PC users who want to reduce overall power consumption and heat dissipation. The processor models will be designated with the "LE" prefix for the Athlon and Sempron families. The X2 family will get the "BE" designation. The new models and pricing per CPU in quantities of 1000 are:
# Athlon X2 BE-2400 (2.3 GHz) - $104
# Athlon LE-1620 (2.4 GHz) - $53
# Athlon LE-1600 (2.2 GHz) - $47
# Sempron LE-1250 (2.2 GHz) - $53
# Sempron LE-1200 (2.1 GHz) - $48
# Sempron LE-1100 (1.9 GHz) - $37
Intel X38 (Code name: Bearlake-X) chipset was launched on October 10th. The X38 chipset will support 800 MHz/1066 MHz/1333MHz FSB, up to DDR3-1333 memory, PCI Express 2.0 (PCIe) and dual PCIe x16 graphics slots, and will also be the first core logic chipset using ISH (integrated heat spreader) technology. A new feature includes the ability to disable USB and SATA ports when they are un-needed.
nVidia G92 (Code name: D8M = Desktop 8 Performance) graphics chip was announced on October 29th (originally November 12th). The G92 will compete against AMD's RV670 as nVidia's performance graphics chip. Performance wise, the G92 should fall between the existing 8800 GTS and 8800 GTX models. G92 supports nVidia PureVideo HD technology, HDCP, DirectX 10, OpenGL 2, PCIe 2.0, 112 stream processors running at 1.5 GHz, and feature a 600 MHz core clock, 900 MHz memory clock (GDDR3), supports up to 1024MB of memory, and has a 256 bit memory interface. The chips are built using TSMC's 65 nm processor which will be a first for nVidia. Graphics cards based on the G92 chip are called GeForce 8800 GT. 8800 GT has a draw of 105 Watts during high loads.
nVidia released the Quadro FX 5600 SDI and Quadro FX 4600 SDI on October 30th. The 5600 and 4600 are aimed at video professional who need to be able to produce video through real-time rendering. Both cards can output SD, HD, and 2K formatting video. Both can contain either 768MB to 1.5GB of frame buffer memory. Cards will retail for between $5999 and $6999.

November

Intel Itanium 9100 series (Code name: Montvale) was released on November 1st. Montvale is a refresh of the existing Itanium lineup and will be released with a slightly faster speed of 1.6 GHz (compared to the existing Itanium's 1.4 GHz). The FSB has been bumped up to 667 MHz from 533 MHz. L3 cache sizes range from 8MB up to 24MB. The Itanium chips gain a new feature called Core Level Lockstep which insures that if the same calculations are done on two cores on any processor in the system, the results match. Demand Based Switching (DBS) has also been added to reduce energy consumption during periods of low utilization. Initial prices will range from $696 to $3,692.
Intel Itanium 9100 series (Code name: Dimona) was released at the same time as the Intel Itanium 9100 (Montvale). Dimona is the dual core version of Montvale.
ATI Radeon HD 2900 GT was released on November 8. The HD 2900 GT contains 700 million transistors and manufactured using an 80 nm process. It also contains a 256-bit 4-channel GDDR memory interface and 240 shader stream processing units. It is DirectX 10 compatible and offers 5.1 surround sound through HDMI. The core frequency is 600 MHz and the memory clocks in at 1600 MHz.
nVidia GeForce 8800 GTS SSC was released on November 8. It is exactly like the GeForce 8800 GTS except it has 112 stream processors instead of 96.
ATI FireStream 9170 was launched on November 9th. FireStream is a general purpose GPU capable of double-precision floating point calculations. The GPU also is parallel processing capable with 320 stream cores, 2GB GDDR3 memory, and PCIe 2.0 x16 compatible. It has 500 GFLOPs of single precision performance.

Intel Penryn was launched on November 12th. Penryn is built on the new 45 nm manufacturing process. Each Penryn die will contain 410 million transistors. Intel has stated that Penryn will increase performance by 20 percent and also have improved power efficiency. Some early benchmarks have shown up to 100% performance increases when the applications were SSE4 optimized. The move to 45 nm also moves Intel's silicon fabrication away from a thinning nitrided silicon dioxide gate insulator to a High-K metal gate thus halting power loss via gate leakage. Penryn is an update to the Core 2 micro-architecture and will initially be marketed for use in servers and high-end workstations. Penryn's features include bigger L2 caches (6MB for dual-core CPUs and 12MB for quad), better cache management, improved power efficiency (C6 "Deep Power Down" state), FSB support up to 1600MHz, 2x faster division operations, SSE4, and Enhanced Dynamic Acceleration technology.

Yorkfield will be the quad-core processor for the Penryn family. The "Extreme Edition", QX9650, is expected to be the only processor available on the November 12th launch day. The other Yorkfields will launch sometime in January 2008, possibly around January 7th at the start of the 2008 International Consumer Electronics Show.

Yorkfield Quad-Core - 1333 MHz FSB
Model	Core	L2	Price
QX9650	3.0 GHz	12MB	$999
Q9950	2.83 GHz	12MB	$530
Q9450	2.66 GHz	12MB	$316
Q9300	2.50 GHz	12MB	$266

Wolfdale are the dual-core variant of the Penryn family of processors. The main difference between the Yorkfield and Wolfdale processors, aside from dual vs. quad core, is the amount of L2 cache. Wolfdale will only have half, 6MB, the L2 cache of Yorkfield. Wolfdale processors will be launched with the remaining Yorkfield processors in January 2008.

Wolfdale Dual-Core - 1333 MHz FSB
Model	Core	L2	Price
E8500	3.16 GHz	6MB	$266
E8400	3.00 GHz	6MB	$183
E8200	2.66 GHz	6MB	$???

Two Intel Xeon (Code name: Harpertown) processors where released on November 12th. The X5472 is a high-end quad-core CPU built using the new 45 nm manufacturing process. It features a 1600 MHZ FSB and 3.0 GHz clock speed with 12MB L2 cache and 120W TDP. The other processor, the X5450, has the same features as the X5472 but with a slower 1333 MHz FSB.
AMD Phenom X4 (Code name: Agena) was released on November 19th. Phenom is the socket AM2 variant of the quad core AMD Opteron (Barcelona) CPU for the desktop market. Like Barcelona, Phenom features 4x 512KB L2 cache, 2MB of shared L3 cache, and HT3 running at 3.6GHz. At launch, the Phenom 9600 and the 9500 will debut with clock speeds of 2.3GHz to 2.2 GHz respectively. Both will have an 89 Watt TDP. Pricing per processor will initially be around $320 and $280 respectively.
AMD Phenom X2 (Code name: Kuma) was released along with the AMD Phenom X4. Phenom X2 is the dual core variant of the Phenom X4. It will feature 2x 512KB L2 and 2MB L3 cache. It will also debut in speeds ranging from 2.0GHz to 2.4GHz.
ATI RV670 was launched on November 19th. The RV670 graphics core was previously known as the Radeon HD 2950 but will be branded the Radeon HD 3800 instead. The RV670 is manufactured using a 55nm process. It will have 320 stream processors with integrated Shader Model 4.1 capability and PCI Express 2.0 support. The RV670 core is capable of DirectX 10.1, bettering its rival the nVidia G92 GPU which only has DirectX 10 support. It also supports CrossFire in dual, triple, and quad card configurations. The previously named PRO, XT, GT, and XTX models will now be replaced with numbers corresponding to the last two digits of the model (i.e. 38XX). The RV670PRO will now be known as the Radeon HD 3850 and have 256MB GDDR3 @ 825 MHz and a core frequency of at least 660 MHz. The RV670XT will be the Radeon HD 3870 and instead of GDDR3 it will have GDDR4 memory clocked at 1.2 GHz and core frequency of 775 MHz. These new model names are to bring the video and CPU lines from AMD in line with the same naming convention.
AMD RD790 chipset was launched along with the AMD Phenom on November 19th. RD790 is AMD's next generation desktop chipset replacing the 580X chipset and will be called the 790-series when it comes to market. Three variants of the chipset will be available: 790FX, 790X, and 790. The 790FX targets the enthusiast market, the 790X targets the performance market, and the 790 is the mainstream variant. The RD790 chipset supports single and dual processor configurations. Processor support includes both older AM2 and newer AM2+ processors as well as the upcoming Phenom X2, X3, and X4 processors. RD790 includes support for PCI-Express 2, HT3.0, DDR2/DDR3, increased SATA ports, eSATA, and support up to 42 PCIe 2.0 lanes. Dual, triple, and quad card Crossfire ("CrossFireX") configurations will also be available on 790FX and 790X motherboards.
nVidia launched the GeForce 8800M GTS and GTX mobile GPU's on November 19th. The 8800M series GPU's will be marketed as enthusiast GPU's in the mobile market. They both will have a 500MHz core frequency, 1600MHz memory frequency, and 512MB memory accessible thru a 256-bit memory interface. The GTS has 64 stream processors and the GTX has 96. Both support DirectX 10 and PureVideo HD. Maximum power usage for the GTX is 37 Watts while the GTS is slightly lower.

December

Microsoft Office 2007 Service Pack 1 (SP1) was released on December 11th. In addition to small bug fixes and security updates, SP1 includes:
- Better support for Visual Studio Tools for Office (VSTO) v3
- Improved documentation for the Object Model
- .NET Framework 3.5 improvements
- Animated text in PowerPoint has been enhanced
- Compatibility fixes for previous Office file formats
- DirectX video overlay changes
nVidia released yet another revision of the 8800 GTS (Code name: D8P) on December 11th. nVidia began using a 65 nm manufacturing process instead of the existing 90 nm for the G80 core. These new GTS cards will have 512MB of memory instead of 640MB like the previous model. The new cards will support PCI Express 2.0, HDMI, and PureVideo Gen2. The new 8800 GTS will have 128 steaming processors, a 970MHz GDDDR memory frequency, 650MHz core frequency, and 140 Watt TDP. Card vendors may release two different versions of the card, one with 112 and another with 128 streaming processors. The card may also support up to 1GB of GDDR3 memory.
nVidia nForce 780 and 750 chipsets were announced on December 17th. Mass availability is not expected until March. The 780i SLI and 750i SLI will be for Intel processors and the 780a SLI and 750a SLI for AMD. The "i" variants will be based on the C72 XE and C72 P. The "a" variants based on the MCP72 XE and MCP72 P. The 780 series will feature three PCIe x16 Gen2 support with SLI and be marketed towards enthusiasts. The 750 will only support two PCIe x8 Gen2 slots. Both chipsets will include nVidia's ESA (Enthusiast System Architecture) for hardware status monitoring.
nVidia C72 chipset was released on December 17th. C72 will be a socket 775 chipset for the new 45nm Intel (Penryn) CPU. The C72 will have a 1333MHz FSB and DDR2-1066MHz. The C72 and C73 chipsets are expected to come in various flavors of SLI sporting 3x16 and 2x8 configurations.
nVidia MCP72 chipset was also released on December 17th; however it will be for newer AM2+ processors. It will come with PCI Express (PCIe) 2.0 and Hyper-Transport (HT) 3.0. Unlike the C72 and C73, the MCP72 will only have included Hybrid SLI. Hybrid SLI enables discrete and integrated graphics to work together allow leading to increased performance. Although it was announced in December, shipping will not commence until sometime in January in order to compete against rival AMD's RD790 chipset.
AMD Phenom 9600 Black Edition was released December 19th. The 9600 Black Edition CPU features a 2.3GHz core frequency, 2MB shared L3 cache, and 512KB L2 cache per core. It also supports DDR2 1066MHz memory just like the regular Phenom 9600 CPU. The difference is Black Edition processors have an unlocked clock multiplier for overclockers.
nVidia C73 will also be released around the same time as the C72. It will be the enthusiast version supporting 1600MHz FSB and DDR3-1600.
nVidia D8M (Desktop 8 Mainstream) (Previous code name: G98) graphics chip is also expected to be released after the G92 on October 29th, most likely in Deceber. D8M is positioned to replace the GeForce 8400 series and should be called the GeForce 8400 GS. The new chip should support PCIe 2.0 and come with 256 MB of 2.5ns GDDR2, a 64-bit memory interface, and built using the new 65 nm process.
nVidia MCP78 is expected to begin mass production in December 2007 and launched in January & February 2008. MCP78 will consist of two new GeForce 8 series IGP (Integrated Graphics Processor) solutions and one without. The "U" and "S" models will be DirectX 10 compatible with a unified memory architecture PureVideo HD technology for high-definition HD DVD and Blu-ray playback. They will also support Hybrid SLI. The "D" model will not have Hybrid SLI or PureVideo support. All of the MCP78 solutions will have one PCIe x16 Gen2 and HT3.0. The "D" model will have a price of around $55 while the "U" has a price around $78.

About techfuzz

Quick Links

Contact