|
|
xtreview is your : Video card - cpu - memory - Hard drive - power supply unit source
|
|
|
POSTER: computer news || GEFORCE GTX 1080 TI AND KABY LAKE-X HAVE DISTINGUISHED THEMSELVES IN UNIGINE HEAVEN XTREME
|
DATE:2017-08-01
|
The American enthusiast Splave at the end of last week updated the absolute world record Unigine Heaven Xtreme, gaining 11 608.83 points DX11 with a video card GeForce GTX 1080 Ti and the central processor Intel Core i7-7740X. The latter was overclocked to 7.0 GHz with liquid nitrogen.
The video card ROG GeForce GTX 1080 Ti STRIX manufactured by Asus, as far as the available screenshot allows, was overclocked to 2190/12008 MHz using liquid nitrogen, or to even higher frequencies, which the author of the experiment was too shy to indicate in the HWBot database. But he explains that a reservoir for liquid nitrogen Der8auer Raptor was used. Thermal Grizzly was used everywhere, and the CPU was able to conquer the 7.1 GHz frequency. Select Rating : 49374 Rating: 10.0/10 (21 votes cast) GeForce GTX 1080 Ti and Kaby Lake-X have distinguished themselves in Unigine Heaven Xtreme |
Add Comment ( 1 )
GEFORCE GTX 1080 TI AND KABY LAKE-X HAVE DISTINGUISHED THEMSELVES IN UNIGINE HEAVEN XTREME |
|
POSTER: computer news || GPU AMD VEGA 10 DESIGN IS A BREAKTHROUGH FOR PRODUCTIVITY GROWTH
|
DATE:2017-08-01
|
The official announcement of video game cards on AMD Vega 10 architecture - Radeon RX Vega 56 and Radeon RX Vega 64 models - lifted the ban on publishing a number of details about the new graphics architecture. Due to leaks, some of the data was previously known, but now we have an official confirmation of all previously received unofficial information.
So, the area of RX Vega graphics processor is 486 mm2, which is 112 mm2 less than the area of the Fiji graphics processor. At the same GPU Vega 10 has 12.5 billion transistors, while in the Fiji Fiji there are "only" 8.9 billion. It is not possible to explain the decrease in the labor GPU area . The previous processor was manufactured using a 28-nm process technology, while the GPU on the Vega 10 architecture is manufactured using the 14LPP process technology - this is the second
generation of the 14-nm process technology. Reducing the geometric dimensions of the transistor has brought its effect - a smaller area was able to squeeze significantly more transistors (in the photo Vega 10 at the top, and Fiji at the bottom).
Another interesting thing is where did the "extra" budget from 3.9 billion transistors go, because the number of pipelines since the Fiji Fiji remained unchanged and equal to 4,096 units? As explained in AMD, the new architecture provides new levels of pipelines. Additional levels allow to "mask" the delay, which allow to significantly increase the speed of the new GPU. Thus, AMD Fiji graphics processors can operate at a frequency of up to 1.05 GHz, and GPUs on AMD Vega
10 are capable of operating at a frequency of up to 1.7 GHz.
As we know, the transfer of production from technological standards of 28 nm to 14 nm gives the possibility of an increase in the clock frequency by approximately 30%. The GPU on the Vega 10 architecture from the transition to the 14LPP process technology was incremented by about 350 MHz. The increase of 400 MHz is the merit of AMD system architects.
It is also officially confirmed that the ratio of FP64 units to FP32 units in the RX Vega GPU is 1/16, as in the case of other (previous) AMD video gaming and professional solutions of the previous generation. Such GPUs are not suitable for high-performance general purpose computing, but can be used in special niches where resources are required to accelerate FP16 and INT8 operations. First of all, it is a newfangled sphere with AI and deep machine learning. Finally, the RX Vega GPU can not
use the ECC data integrity check feature, although the processor itself can work with HBM2 memory, which is initially armed with EEC support. In the processor for gaming adapters there is no necessary wiring.
As for the use of HBM2 memory, there is one more interesting detail. According to our Japanese colleagues, gaming adapters carry two four-layer HBM2 chips, each of which has a capacity of 4 GB (the total amount of memory on the GPU RX Vega substrate is 8 GB). This means that each HBM2 stack in the RX Vega GPU consists of 8-Gbit memory chips. Potentially HBM2 stack can consist of 8 crystals. Select Rating : 49373 Rating: 10.0/10 (19 votes cast) GPU AMD Vega 10 design is a breakthrough for productivity growth |
Add Comment ( 1 )
GPU AMD VEGA 10 DESIGN IS A BREAKTHROUGH FOR PRODUCTIVITY GROWTH |
|
POSTER: computer news || SAMSUNG MANAGED TO BECOME THE LARGEST MANUFACTURER OF SEMICONDUCTOR PRODUCTS
|
DATE:2017-08-01
|
Experts have long predicted that the rise in prices for memory will allow Samsung Electronics in the second quarter to become the largest semiconductor manufacturer in the world in monetary terms. According to the EE Times website with reference to the estimates of IC Insights, in the first half of the year Samsung Electronics really managed to bypass Intel in terms of revenue, and become the largest player in the semiconductor market. Intel did not release the palm from its hands since 1992, but
now IC Insights experts are sure that even the unstable nature of the memory market will not prevent Samsung Electronics from regularly heading this rating. The South Korean company's business can grow faster than the market, and Intel's likely figure will hardly exceed 5-6% per year, as the industry average.
An excursion into the history of statistics in this direction allows us to conclude that Intel regularly increases its revenue, but Samsung Electronics is able to develop at a faster pace. In general, the participants in the "top ten" account for 58% of the semiconductor market revenue, so the trend towards consolidation and consolidation is clearly traced.
Select Rating : 49372 Rating: 10.0/10 (20 votes cast) Samsung managed to become the largest manufacturer of semiconductor products |
Add Comment ( 1 )
SAMSUNG MANAGED TO BECOME THE LARGEST MANUFACTURER OF SEMICONDUCTOR PRODUCTS |
|
POSTER: computer news || GPU AMD VEGA 10 DESIGN IS A BREAKTHROUGH FOR PRODUCTIVITY GROWTH
|
DATE:2017-08-01
|
The official announcement of video game cards on AMD Vega 10 architecture - Radeon RX Vega 56 and Radeon RX Vega 64 models - lifted the ban on publishing a number of details about the new graphics architecture. Due to leaks, some of the data was previously known, but now we have an official confirmation of all previously received unofficial information.
So, the area of RX Vega graphics processor is 486 mm2, which is 112 mm2 less than the area of the Fiji graphics processor. At the same GPU Vega 10 has 12.5 billion transistors, while in the Fiji Fiji there are "only" 8.9 billion. It is not possible to explain the decrease in the labor GPU area . The previous processor was manufactured using a 28-nm process technology, while the GPU on the Vega 10 architecture is manufactured using the 14LPP process technology - this is the second
generation of the 14-nm process technology. Reducing the geometric dimensions of the transistor has brought its effect - a smaller area was able to squeeze significantly more transistors (in the photo Vega 10 at the top, and Fiji at the bottom).
Another interesting thing is where did the "extra" budget from 3.9 billion transistors go, because the number of pipelines since the Fiji Fiji remained unchanged and equal to 4,096 units? As explained in AMD, the new architecture provides new levels of pipelines. Additional levels allow to "mask" the delay, which allow to significantly increase the speed of the new GPU. Thus, AMD Fiji graphics processors can operate at a frequency of up to 1.05 GHz, and GPUs on AMD Vega
10 are capable of operating at a frequency of up to 1.7 GHz.
As we know, the transfer of production from technological standards of 28 nm to 14 nm gives the possibility of an increase in the clock frequency by approximately 30%. The GPU on the Vega 10 architecture from the transition to the 14LPP process technology was incremented by about 350 MHz. The increase of 400 MHz is the merit of AMD system architects.
It is also officially confirmed that the ratio of FP64 units to FP32 units in the RX Vega GPU is 1/16, as in the case of other (previous) AMD video gaming and professional solutions of the previous generation. Such GPUs are not suitable for high-performance general purpose computing, but can be used in special niches where resources are required to accelerate FP16 and INT8 operations. First of all, it is a newfangled sphere with AI and deep machine learning. Finally, the RX Vega GPU can not
use the ECC data integrity check feature, although the processor itself can work with HBM2 memory, which is initially armed with EEC support. In the processor for gaming adapters there is no necessary wiring.
As for the use of HBM2 memory, there is one more interesting detail. According to our Japanese colleagues, gaming adapters carry two four-layer HBM2 chips, each of which has a capacity of 4 GB (the total amount of memory on the GPU RX Vega substrate is 8 GB). This means that each HBM2 stack in the RX Vega GPU consists of 8-Gbit memory chips. Potentially HBM2 stack can consist of 8 crystals. Select Rating : 49371 Rating: 10.0/10 (23 votes cast) GPU AMD Vega 10 design is a breakthrough for productivity growth |
Add Comment ( 1 )
GPU AMD VEGA 10 DESIGN IS A BREAKTHROUGH FOR PRODUCTIVITY GROWTH |
|
|
xtreview is your : Video card - cpu - memory - Hard drive - power supply unit source
|
|
|
| |