GALAX HOF Pro PCIe 4.0 NVMe review
How to: Firmware Update the AMD Radeon RX 5600 XT
MSI Radeon RX 5600 XT Gaming Z review
Sapphire Pulse Radeon RX 5600 XT OC review
ASUS STRIX Radeon RX 5600 XT TOP review
Gigabyte Radeon RX 5600 XT Gaming OC review
Arctic Liquid Freezer II 280 liquid cooler review
Promo: Windows 10 Pro licenses for under 10 USD
Team Group MP33 NVMe 512 GB SSD Review
ADATA SE800 1TB Portable SSD review
Intel preps 16-core Xeon D-1571 processor at 45 Watt
Though intended for micro-servers, Intel is prepping a Xeon D-1571 processor rated at 45 Watt, the beauty of it all, it has 16 Broadwell CPU coresand 32 logical CPUs with Hyper Threading. They run at 1.30 GHz clock speed and are tied to a dual-channel DDR4 memory controller that will support up to 128 GB ECC DDR4 memory.
Particularly, Xeon D-1500 series makes use of Broadwell 14nm microarchitecture, which supports all latest x86 extensions. Currently, the fastest in the series is the D-1541, which has 8 CPU cores and 12 MB L3 cache.
As CPU world reports D-1571 SoC will double the number of cores to 16, albeit running at 35% lower clock rate. The D-1571 will also have twice larger level 3 cache. 16 CPU cores, 1.3 GHz clock speed, 24 MB last level cache, 45 Watt TDP and support for Hyper-threading feature, which allows the CPU to run 32 threads at once. The SoC is likely to support Turbo Boost technology, but we do not have information on the maximum boost speed yet. The official price of the D-1571 will be $1222.
Intel preps SSD based on 3D Xpoint Memory and it is FAST - 08/19/2015 08:59 AM
This development is very interesting as it can greatly increase R/W IOPS performance with a queue depth of just 1. 3D Xpoint memory technology is the successor to standard NAND flash memory with an E...
Intel Preps for 4K with Intel Iris Graphics 6100 - 02/26/2015 10:15 AM
The Intel Iris Graphics 6100 (GT3) is an integrated Broadwell GPU revealed in Q1 2015. The Broadwell-GT3 silicon, Iris 6100 will get 48 execution units, a 300 to 1100 MHz dynamic GPU core frequency...
Samsung Chromebook 2 Available with Intel Processor - 10/20/2014 08:42 AM
Samsung, the leading Chromebook brand, today announced availability of the new, 11-inch Chromebook 2 with an Intel Celeron (N2840) processor. The laptop is available for pre-order now for an MSRP of $...
Intel Processor Roadmap for 2014 hits the web - 02/18/2014 01:32 PM
Two slides from Intel have appeared on the web today, Vr-Zone listed the processor plans for the remainder of the year. The first slide is a generic roadmap that shows what Intel has in store for its...
Intel preps 165W high-performance CPUs - 06/04/2012 09:19 AM
Word out on the street is that that Intel plans to trade TDP for performance. Over the Ivy Bridge and Haswell generations, the chip giant is anticipated to roll out high-end workstation parts with a T...
vbetts
Moderator
Posts: 14764
Joined: 2006-07-04
Moderator
Posts: 14764
Joined: 2006-07-04
#5232589 Posted on: 02/16/2016 01:50 PM
...Next CPU in Xbox two....
Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
...Next CPU in Xbox two....
Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
Undying
Senior Member
Posts: 12311
Joined: 2008-08-28
Senior Member
Posts: 12311
Joined: 2008-08-28
#5232590 Posted on: 02/16/2016 02:02 PM
Indeed. This thing will be destroyed by a simple 6600k when it comes to games.
It's cpu's like this that make you wish games were more multithreaded.
Indeed. This thing will be destroyed by a simple 6600k when it comes to games.
cyclone3d
Senior Member
Posts: 419
Joined: 2006-11-10
Senior Member
Posts: 419
Joined: 2006-11-10
#5232682 Posted on: 02/16/2016 05:40 PM
...Next CPU in Xbox two.... Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
Indeed. This thing will be destroyed by a simple 6600k when it comes to games.
Yes, single threaded performance is king for now for most everything.
However, some games and other applications would do better with more threads.
Massive RTS games for example. Now that is something that could be made to use a ton of threads.
Most games cannot be more threaded than they currently are.. or at least it would not be worth the trouble for the minuscule amount of performance gained in certain situations.
If you are playing a game and don't have one core that is pegged out for usage, then more threading will probably not help. And even then, sometimes that thread that is pegging out usage has to run linearly.
the only other option is that the game was coded poorly and the pegged out thread is just wasting CPU cycles.
...Next CPU in Xbox two....
Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
Indeed. This thing will be destroyed by a simple 6600k when it comes to games.
Yes, single threaded performance is king for now for most everything.
However, some games and other applications would do better with more threads.
Massive RTS games for example. Now that is something that could be made to use a ton of threads.
Most games cannot be more threaded than they currently are.. or at least it would not be worth the trouble for the minuscule amount of performance gained in certain situations.
If you are playing a game and don't have one core that is pegged out for usage, then more threading will probably not help. And even then, sometimes that thread that is pegging out usage has to run linearly.
the only other option is that the game was coded poorly and the pegged out thread is just wasting CPU cycles.
Ven0m
Senior Member
Posts: 1764
Joined: 2005-08-12
Senior Member
Posts: 1764
Joined: 2005-08-12
#5232758 Posted on: 02/16/2016 08:49 PM
It's not only about "poor coding". Achieving good multi-core performance is especially hard in multiplayer games.
Basically, for 60fps game you have 16ms to prepare entire frame. 120Hz - 8ms. It's tough. You have to calculate all the physics, game logic, network code, audio, and graphics in so short time. Work partitioning, distribution, asynchronous computation, gathering results and composition have to complete in this tiny fraction of second. Of course, some of the computations have value for longer time than 1 frame, eg. some AI, pathing, and general forecasts, but paired with network synchronization, it's super-challenging. On top of that, synchronization between cores (making one core aware that other has finished and flushing the data) takes a lot of time.
Distribution of non-time-critical workload is doable and while not trivial, it's well-described. I wouldn't expect distributing of tasks like engineering computations or rendering across many cores and machines to be that difficult for an experienced software developer. But loading 8+ cores effectively with real-time flow of multiple data streams that depend on each other, is exceptionally tough.
This makes me think that:
a) Games will be optimized for max 4 cores for quite long time
b) Games will be capped to 60 or even 30 fps to give more time to calculate the frame
It's not only about "poor coding". Achieving good multi-core performance is especially hard in multiplayer games.
Basically, for 60fps game you have 16ms to prepare entire frame. 120Hz - 8ms. It's tough. You have to calculate all the physics, game logic, network code, audio, and graphics in so short time. Work partitioning, distribution, asynchronous computation, gathering results and composition have to complete in this tiny fraction of second. Of course, some of the computations have value for longer time than 1 frame, eg. some AI, pathing, and general forecasts, but paired with network synchronization, it's super-challenging. On top of that, synchronization between cores (making one core aware that other has finished and flushing the data) takes a lot of time.
Distribution of non-time-critical workload is doable and while not trivial, it's well-described. I wouldn't expect distributing of tasks like engineering computations or rendering across many cores and machines to be that difficult for an experienced software developer. But loading 8+ cores effectively with real-time flow of multiple data streams that depend on each other, is exceptionally tough.
This makes me think that:
a) Games will be optimized for max 4 cores for quite long time
b) Games will be capped to 60 or even 30 fps to give more time to calculate the frame
quadibloc
Junior Member
Posts: 1
Joined: 2016-02-16
Junior Member
Posts: 1
Joined: 2016-02-16
#5232833 Posted on: 02/16/2016 11:36 PM
Given that my gaming laptop has an Intel Core i7 with a TDP of 45 watts, this 16-core processor, apparently, could actually be put in a laptop! That, of course, would be very impressive, even if 1.3 GHz isn't as fast as some chips with fewer cores can go.
Given that my gaming laptop has an Intel Core i7 with a TDP of 45 watts, this 16-core processor, apparently, could actually be put in a laptop! That, of course, would be very impressive, even if 1.3 GHz isn't as fast as some chips with fewer cores can go.
Darkest
Senior Member
Posts: 10123
Joined: 2003-03-25
Senior Member
Posts: 10123
Joined: 2003-03-25
#5232836 Posted on: 02/16/2016 11:50 PM
...Next CPU in Xbox two.... Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
Hah.
Makes me think of when everyone was buying Opertons back in the day, less due to core count and more for CPU type that is.
...Next CPU in Xbox two....
Joking of course.
It's cpu's like this that make you wish games were more multithreaded.
Hah.
Makes me think of when everyone was buying Opertons back in the day, less due to core count and more for CPU type that is.
RzrTrek
Senior Member
Posts: 2358
Joined: 2012-04-16
Senior Member
Posts: 2358
Joined: 2012-04-16
#5232844 Posted on: 02/17/2016 12:04 AM
16 x 1.3GHz?
16 x 1.3GHz?
southamptonfc
Senior Member
Posts: 1810
Joined: 2006-04-10
Senior Member
Posts: 1810
Joined: 2006-04-10
#5232868 Posted on: 02/17/2016 01:14 AM
This will make an awesome low power VM server chip. 45W, wow.
This will make an awesome low power VM server chip. 45W, wow.
vbetts
Moderator
Posts: 14764
Joined: 2006-07-04
Moderator
Posts: 14764
Joined: 2006-07-04
#5232902 Posted on: 02/17/2016 03:31 AM
There was a quad core 15w Xeon released not too log ago as well...
There was a quad core 15w Xeon released not too log ago as well...
dwiewolverine
Senior Member
Posts: 163
Joined: 2015-10-20
Senior Member
Posts: 163
Joined: 2015-10-20
#5232932 Posted on: 02/17/2016 05:13 AM
hell yeahh..another low power procie from BLUE side..
hell yeahh..another low power procie from BLUE side..
BedantP
Senior Member
Posts: 220
Joined: 2013-08-31
Senior Member
Posts: 220
Joined: 2013-08-31
#5232965 Posted on: 02/17/2016 08:48 AM
It's not only about "poor coding". Achieving good multi-core performance is especially hard in multiplayer games.
Basically, for 60fps game you have 16ms to prepare entire frame. 120Hz - 8ms. It's tough. You have to calculate all the physics, game logic, network code, audio, and graphics in so short time. Work partitioning, distribution, asynchronous computation, gathering results and composition have to complete in this tiny fraction of second. Of course, some of the computations have value for longer time than 1 frame, eg. some AI, pathing, and general forecasts, but paired with network synchronization, it's super-challenging. On top of that, synchronization between cores (making one core aware that other has finished and flushing the data) takes a lot of time.
Distribution of non-time-critical workload is doable and while not trivial, it's well-described. I wouldn't expect distributing of tasks like engineering computations or rendering across many cores and machines to be that difficult for an experienced software developer. But loading 8+ cores effectively with real-time flow of multiple data streams that depend on each other, is exceptionally tough.
This makes me think that:
a) Games will be optimized for max 4 cores for quite long time
b) Games will be capped to 60 or even 30 fps to give more time to calculate the frame
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
It's not only about "poor coding". Achieving good multi-core performance is especially hard in multiplayer games.
Basically, for 60fps game you have 16ms to prepare entire frame. 120Hz - 8ms. It's tough. You have to calculate all the physics, game logic, network code, audio, and graphics in so short time. Work partitioning, distribution, asynchronous computation, gathering results and composition have to complete in this tiny fraction of second. Of course, some of the computations have value for longer time than 1 frame, eg. some AI, pathing, and general forecasts, but paired with network synchronization, it's super-challenging. On top of that, synchronization between cores (making one core aware that other has finished and flushing the data) takes a lot of time.
Distribution of non-time-critical workload is doable and while not trivial, it's well-described. I wouldn't expect distributing of tasks like engineering computations or rendering across many cores and machines to be that difficult for an experienced software developer. But loading 8+ cores effectively with real-time flow of multiple data streams that depend on each other, is exceptionally tough.
This makes me think that:
a) Games will be optimized for max 4 cores for quite long time
b) Games will be capped to 60 or even 30 fps to give more time to calculate the frame
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
southamptonfc
Senior Member
Posts: 1810
Joined: 2006-04-10
Senior Member
Posts: 1810
Joined: 2006-04-10
#5233026 Posted on: 02/17/2016 11:04 AM
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
It's not that easy. Architecture changes don't automatically solve multi-threading design issues. It's neither easy or always possible to actually use all the threads available in a system, certain applications lend themselves better to multi-threading than others.
Anyway, how many games are actually CPU bound these days? Not a lot. If you're running >= 1080p, any reasonable quad core from the few years is sufficient.
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
It's not that easy. Architecture changes don't automatically solve multi-threading design issues. It's neither easy or always possible to actually use all the threads available in a system, certain applications lend themselves better to multi-threading than others.
Anyway, how many games are actually CPU bound these days? Not a lot. If you're running >= 1080p, any reasonable quad core from the few years is sufficient.
xIcarus
Senior Member
Posts: 941
Joined: 2010-08-24
Senior Member
Posts: 941
Joined: 2010-08-24
#5233104 Posted on: 02/17/2016 01:46 PM
It's not that easy. Architecture changes don't automatically solve multi-threading design issues. It's neither easy or always possible to actually use all the threads available in a system, certain applications lend themselves better to multi-threading than others.
Anyway, how many games are actually CPU bound these days? Not a lot. If you're running >= 1080p, any reasonable quad core from the few years is sufficient.
Quite a number of games in fact, especially games where there's a lot of stuff happening on the screen aka a lot of units NPCs etc. See below.
Yes, single threaded performance is king for now for most everything.
However, some games and other applications would do better with more threads.
Massive RTS games for example. Now that is something that could be made to use a ton of threads.
Most games cannot be more threaded than they currently are.. or at least it would not be worth the trouble for the minuscule amount of performance gained in certain situations.
If you are playing a game and don't have one core that is pegged out for usage, then more threading will probably not help. And even then, sometimes that thread that is pegging out usage has to run linearly.
the only other option is that the game was coded poorly and the pegged out thread is just wasting CPU cycles.
Not with DX11. Massive RTS and MMO games suffer the most because of the large amount of draw calls present. Considering that in DX11 communication between the CPU and GPU is done on 1 thread, draw calls are limited by design.
This is where DX12 and Vulkan will come in. I'm sick and tired of SC2 having deplorable FPS every time there's a huge fight on the screen. Especially since I play a lot of 3v3 and 4v4 arcade games. I still have a screenshot on my desktop, look at this: http://s9.postimg.org/6l6564fj3/fuuuuck.png
2 FPS. I hope Blizzard will add Vulkan or DX12 support.
It's not that easy. Architecture changes don't automatically solve multi-threading design issues. It's neither easy or always possible to actually use all the threads available in a system, certain applications lend themselves better to multi-threading than others.
Anyway, how many games are actually CPU bound these days? Not a lot. If you're running >= 1080p, any reasonable quad core from the few years is sufficient.
Quite a number of games in fact, especially games where there's a lot of stuff happening on the screen aka a lot of units NPCs etc. See below.
Yes, single threaded performance is king for now for most everything.
However, some games and other applications would do better with more threads.
Massive RTS games for example. Now that is something that could be made to use a ton of threads.
Most games cannot be more threaded than they currently are.. or at least it would not be worth the trouble for the minuscule amount of performance gained in certain situations.
If you are playing a game and don't have one core that is pegged out for usage, then more threading will probably not help. And even then, sometimes that thread that is pegging out usage has to run linearly.
the only other option is that the game was coded poorly and the pegged out thread is just wasting CPU cycles.
Not with DX11. Massive RTS and MMO games suffer the most because of the large amount of draw calls present. Considering that in DX11 communication between the CPU and GPU is done on 1 thread, draw calls are limited by design.
This is where DX12 and Vulkan will come in. I'm sick and tired of SC2 having deplorable FPS every time there's a huge fight on the screen. Especially since I play a lot of 3v3 and 4v4 arcade games. I still have a screenshot on my desktop, look at this: http://s9.postimg.org/6l6564fj3/fuuuuck.png
2 FPS. I hope Blizzard will add Vulkan or DX12 support.
Ven0m
Senior Member
Posts: 1764
Joined: 2005-08-12
Senior Member
Posts: 1764
Joined: 2005-08-12
#5233265 Posted on: 02/17/2016 06:42 PM
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
A single core being used should be considered a defect today. Loading 4 cores to 50%+ average should be a standard in games. But getting to 16 or 32 cores loaded at 80%+ in a game is borderline impossible now, at least with high frame rate.
Decreasing sync latency will help a lot. Even now, it's possible to benefit from distributing computations across cores, but it's much more demanding to do that than to assign each core a dedicated task. It's done today, as it's easier - you don't need that good devs to achieve it. And it makes me angry that game like Diablo 3 can slow down on a fast computer, while the graphics card is barely loaded, and all cores but 1 or 2 are idling.
On the other hand, such CPUs are great for servers, where each core can work for a different user, with close to no dependencies between them and much greater delay tolerance.
I guess that it will change once low-speed 8/16 core CPUs become a new default for low/mid-end computers and there will be no other way to scale speed than to learn how to distribute the computations.
So this leaves us with only one way to improve which is better input-per-core?
New architectures can solve all those latency issues you're talking about.
It's just the devs that code poorly and only 1 core is used and others are at rest.
A single core being used should be considered a defect today. Loading 4 cores to 50%+ average should be a standard in games. But getting to 16 or 32 cores loaded at 80%+ in a game is borderline impossible now, at least with high frame rate.
Decreasing sync latency will help a lot. Even now, it's possible to benefit from distributing computations across cores, but it's much more demanding to do that than to assign each core a dedicated task. It's done today, as it's easier - you don't need that good devs to achieve it. And it makes me angry that game like Diablo 3 can slow down on a fast computer, while the graphics card is barely loaded, and all cores but 1 or 2 are idling.
On the other hand, such CPUs are great for servers, where each core can work for a different user, with close to no dependencies between them and much greater delay tolerance.
I guess that it will change once low-speed 8/16 core CPUs become a new default for low/mid-end computers and there will be no other way to scale speed than to learn how to distribute the computations.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 2254
Joined: 2007-05-31
45watt, dual chanel so it is low cost motherboard (for pro of course lol), it start to make shadow to Atom config for files server and micro server...