AMD or Intel as well as the memory brand - page 78
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Who's had it, Pete? I'm still looking at the 1156 socket, don't want to go wrong with the motherboard.
http://www.ixbt.com/news/index.php?news_id=125205&page=1#news125205
Are you (or are you? I can't remember anymore))) referring to burned boards due to loose fit between the stone and socket contacts? Yes, it happened, but only at one socket supplier.
But yeah - I couldn't sleep when I read it! What if I think I might get such a thing! )))
You were drinking vodka...
;)))
What do you mean burnt?
Like a defect in the socket or the assembler?
No! I'm interested in something else, a strategic one, so as not to make a mistake with the platform.
Now I'll save a couple of matches and buy a box of matches which won't burn after half a year.
That's what I meant by "Intel's playing a joke again" ...
Well, how can they not burn when Intel releases the i3, i5, i7 series, aimed at the most mass segment - housewives. They will still be on fire in five years' time.
Well, how can they not burn when Intel releases the i3, i5, i7 series, aimed at the most mass segment - housewives. They will still be on fire in five years' time.
Again, it's all about the choice between 1156 and 1366.
that's it...
There's a difference between them, isn't there? Surely...
Soon everyone will start producing say 1156 and 1366 on a "residual principle",
and in two years you will have to buy the motherboard or 1366 CPU that you have (give)
not the choice which is rich among 1156.
(or vice versa.)
AMD to release Thuban six-core processors in second quarter 2010
Recent AMD roadmap leaks let us know that AMD will ship 45nm six-core Thuban desktop processors in 2010 that will retain Socket AM2+ and Socket AM3 compatibility. One year later, 32nm Zambezi processors with more than four cores and Socket AM3 second revision design should be released. Colleagues from X-bit labs site took the trouble to find out when 45 nm six-core Thuban processors will be available. It turns out that their announcement is now scheduled for the second quarter of 2010, not the third as previously thought. A new Leo platform will hit the market in May 2010, which will include Thuban processors, AMD 890FX chipset and SB850 southbridge. The latter will bring long promised support for SATA-600 and fourteen USB 2.0 ports.
Sources suggest that the announcement of the six-core Phenom II X6 processors has been postponed by a quarter to compete with the six-core Core i9 (Gulftown) processors that Intel will introduce in the second quarter of 2010. The Thuban processors are expected to get 6 x 512Kb of Layer 2 cache and 6MB of Layer 3 cache. Memory controller will support dual-link DDR3-1333, while power subsystem will enable these processors to run on Socket AM2+ and Socket AM3 socket. Heat dissipation level and frequency of Thuban CPUs has not yet been determined.
http://www.overclockers.ru/hardnews/34637.shtml
LGA 1156 design flaw threatens CPU damage
The arrival of new platforms on the market is often accompanied by various problems that could not be detected during development. AnandTech journalists may have identified one of the first serious problems with the new LGA 1156 socket, which debuted in early September.While doing extreme overclocking, their test Core i7-870 processor failed. The cause was thermal damage to the processor and the socket: An investigation revealed that a design flaw in the socket was to blame. The pins were not at the same height and some of them did not reach the contact pads of the processor. When the processor is inserted into the socket, the pins leave traces on the contact pads. The Foxconn connector did not leave such marks on a significant portion of the pads, indicating poor pinning:
But back to the reasons that caused the damage. As we've already mentioned, some of the pins weren't making a good contact with the CPU, which in turn led to a decrease in the effective number of pins responsible for powering the processor. As you know, under load the processor starts to draw more current, which in turn heats up the conductors, i.e. the pins of the connector. Reducing the effective number of conductors greatly increases the specific current that passes through the pins used, and since modern processors consume tens of amps, excessive heating of the pins is not surprising. To make the situation worse the number of pads dedicated to power supply for the processor is almost one and a half times less in LGA 1156 than in LGA 1366, but the overclocking current consumption of these processors is comparable. By the way, the problem is unlikely to occur under normal operating conditions, but overclocking significantly increases the probability of its occurrence. Therefore, the users of new platforms should pay attention whether the pins remain on all pins pads of processors.
We'll keep an eye on the situation and wait for official statements from the manufacturers, but for now we can only wish to be vigilant when overclocking the new platform.
http://www.overclockers.ru/hardnews/34725.shtml
Found as a potential answer to the question:
Intel had no plans to significantly reduce the price of processors for the Socket 1366 platform. Instead, a simpler and cheaper alternative was being developed in the form of Socket 1156 platform, aimed at the mainstream market segment, but like Socket 1366, using Nehalem architecture at its core. First production samples of CPUs and motherboards for the new platform were ready a few months ago, but the start of sales was delayed until September, probably to avoid interfering with sales of Socket 1366 platform.
The X 58 LGA 1366
X58 IOH North Bridge was connected to the CPUs via a fast QPI bus with a bandwidth of 25.6 GB/s. This was necessary in order to pass through the huge amount of data flowing between the processor and the video cards. To communicate with the graphics cards, the north bridge had support for 36 PCI-Express 2.0 lanes. This made it possible to use up to four graphics cards simultaneously in a single system. Each card was allocated 16 lanes when connecting one or two cards, in case of three cards they worked as x16+x8+x8, in case of four - x16+x8+x8+x4. A DMI (Direct Media Interface) bus with 2 GB/s bandwidth was used for communication between the server bridge and the South Bridge.
P55 LGA 1156
The most significant change is the complete absence of a north bridge on the motherboard. "The Intel P55 Chipset consists of only one Platform Controller Hub (PCH) South Bridge and the functionality of the North Bridge has been completely moved to the processor. The graphics cards are now connected directly to the CPU which only supports 16 PCI-Express 2.0 lanes, more than half of the Intel X58. Still it is a chipset for mainstream segment, not for building high-end systems with large number of graphics cards, so it had to sacrifice PCI-Express lanes to reduce system cost for Socket 1156. It is possible to connect two graphics cards to the processor, but in this case they will divide the lines equally between them, i.e. operate according to scheme 8+8. Of course, this would result in some performance degradation of the video subsystem, but this option is quite usable as long as two GPU graphics cards are not combined for this purpose.
Many motherboard manufacturers are going to release solutions on the Intel P55 chipset with the number of PCI-E slots from three or more (for example, the board Asus P7P55 WS Supercomputer has as many as five). Of course, you won't be able to connect more than two graphics cards directly to the processor, as there simply aren't enough PCI-E lines for that. Alternatively, one video card can be connected via the DMI bus connecting the CPU and South Bridge. But the DMI bus bandwidth is several times lower than the QPI bus used by the Intel X58. It is clearly not sufficient for use with powerful graphics cards. The QPI bus is still in the processor, only now it does not protrude beyond the processor. Additionally the DMI bus is actively used by other "traffic consumers" in the CPU <->South Bridge area such as the disk subsystem, embedded audio and Ethernet network interfaces. Connecting another, and such a large, consumer such as a graphics card to them could have a negative impact on the speed of all of these subsystems.
The only way to implement a large number of PCI-E slots so far is to install additional PCI-E switch chips on the motherboards (e.g. NVIDIA NF200 or Lucid Hydra), but this too is far from an ideal solution. A switch merely divides the same connected lines between two video cards and allows them to communicate with each other, without using an "external" channel. But both video cards will still not be able to use the "external" channel at full x16 speed at the same time. To draw an analogy, one could compare a PCI-E line switch to a router with two home computers and one external channel to the Internet for two, hard-limited by speed. However, it is worth noting that motherboards with a large number of PCI-E slots, albeit limited in the number of lines, have their own specific applications - GPU calculations, e.g. for distributed computing (folding).
I think I messed up the link in the last post about faulty sockets. Here is the correct one.
The most important thing, apart from what Olga kindly posted, is in the name of the company that is defecting. Here is an excerpt from the post:
Процессорные разъемы указанного типа для системных плат выпускают компании Foxconn, Tyco AMP и LOTES. Сравнение разных разъемов позволило установить — проблемным является разъем производства Foxconn. Разъемы двух других марок, образно говоря, находятся с процессорами в тесном контакте. По данным источника, разъемы Tyco AMP и LOTES (нижний снимок) используются в системных платах производства EVGA, DFI и MSI.
Strange...
Foxconn has always been known for a good product.
Maybe it was "Foxconn" ;)
from Goduras' central underground shop?
Holy shit!!!
Before I could guess the cause - overheating - this is exactly what it is at the very beginning of the article.
Источник столкнулся с такой ситуацией, когда в ходе экспериментов с экстремальным «разгоном» оказались повреждены все использованные системные платы на чипсете P55 и два весьма недешевых процессора Core i7-870. Как оказалось, причиной стал перегрев и вызванное им разрушение разъема и контактного поля процессора.
Of course, you can drive anything under the bridge...
)))
you should take the 1156 as a cheap solution and at the same time it is an i7.
it is unwise to plan to use a CPU for more than two years if you need a powerful platform.