Will 2.3Ghz Broadwell be enough for gaming?

TheLAWNoob

Limp Gawd
Joined
Jan 10, 2016
Messages
330
Currently have a 4.5Ghz i5 6600K. I got it for DayZ, but now I moved onto better optimized games.

I have my eyes on a $400 16 core 2.3Ghz (all core turbo) Broardwell CPU. 1P mobo will be $300 after tax, and 2P will be $500 after tax. 1P setup should match the i7 6950X in multithread.

Will 2.3Ghz Broardwell be good enough for most of the recent games? I have a 60Hz 4k Freesync monitor.

I'm considering the switch because I might need to render Solidworks videos for next September.

Also, will a 2P system be worth it for the bragging rights? I will be working as an intern full time for 8 month starting from Januaray, and I only need to pay for gas and auto insurance.
 
I run a 1p 2.4GHz 14- core haswell system, and there are times when I can feel the CPU bottlenecking some games. DOOM 2016 still can run at 60FPS, same with GTA and Serious Sam 3.
 
It really depends on the games you play but in general, your 4.5 Ghz 6600k will win easily, sorry.
 
It really depends on the games you play but in general, your 4.5 Ghz 6600k will win easily, sorry.

Of course the i5 will be faster in games.

I only wanted to know if a 2.3Ghz Broadwell can get about 60fps in most recent games.
 
What's your GPU?

I think only a Titan X Pascal will be able to reliably play at 4k. But the good news is at that resolution, your CPU shouldn't be a bottleneck.
 
Of course the i5 will be faster in games.

I only wanted to know if a 2.3Ghz Broadwell can get about 60fps in most recent games.

The short answer is no. Long answer depends a lot on the game.
 
Of course the i5 will be faster in games.

I only wanted to know if a 2.3Ghz Broadwell can get about 60fps in most recent games.
it all depends on how well threaded the game is. It would seriously have to be a major block buster game with BF/COD. 99% of all games will run like total shit on that. Some might work but 2.3GHz is ridiculously low.

What will be bad is the minimum frame rate will be the worst....aka stutters
 
it all depends on how well threaded the game is. It would seriously have to be a major block buster game with BF/COD. 99% of all games will run like total shit on that. Some might work but 2.3GHz is ridiculously low.

What will be bad is the minimum frame rate will be the worst....aka stutters
BTW linus did a video of a multi core xeon with 4 GPUs to make a multi gaming rig. It was an interesting video.

I am hoping AMD enterprise chips are unlocked because i would love to buy a 16 core unlocked chip.
 
BTW linus did a video of a multi core xeon with 4 GPUs to make a multi gaming rig. It was an interesting video.

I am hoping AMD enterprise chips are unlocked because i would love to buy a 16 core unlocked chip.

Dont expect server chips to be unlocked at all. It would be considered a huge bug if they was due to the locations they are to be used.

The 16 core is an MCM design, so you need to lock a game to one set of cores. Whenever windows jumps the threads between dies you are going to be penalized hard. Even on the bolted together SoCs in the consoles its a ~200cycle penalty. This one will only be much higher. Gaming on multi socket systems is just plain bad.
 
Dont expect server chips to be unlocked at all. It would be considered a huge bug if they was due to the locations they are to be used.

The 16 core is an MCM design, so you need to lock a game to one set of cores. Whenever windows jumps the threads between dies you are going to be penalized hard. Even on the bolted together SoCs in the consoles its a ~200cycle penalty. This one will only be much higher. Gaming on multi socket systems is just plain bad.
Intel sold some Xeons unlocked (see my sig)...sadly never the high core ones. AMD could get a lot of sales if they unlocked them.

and no it isn't a bug...they (Intel) just don't unlock them because of profits. Doesn't mean AMD can't or wont

If they sold a 12-16 core chip unlocked i would sell my 1650v3 in a heart beat and buy that 16 core.

I don't see why AMD wouldn't sell an unlocked 16 core when Intel has a 10 core. AMD could get mad amount of business.


Wtf are you talking about multi SoC/ socket system? Linus used a high core CPU not mutli socket system. How about you watch video before talking out of your ass.

are you talking about this?
http://www.anandtech.com/show/10158/the-intel-xeon-e5-v4-review/2

vs this

http://www.anandtech.com/show/10337...x-6900k-6850k-and-6800k-tested-up-to-10-cores

I personally dont know anything about the delay moving from set to set in a CPU but internally in a CPU vs dual socket system is vastly different.

If you throw some 200 cycle bullshit number up back it up

I have never heard of thus penalty on xeons.
 
Last edited:
Solidworks states their application is linear and relies heavily on single threaded performance for most workloads. Look at the system requirements for your apps - this is from a research paper on the solidworks site and a blog site about solidworks:


Processor

Your workstation’s engine is its processor and you only have one chance to get it specified properly. Since the main SOLIDWORKS application is “lightly threaded” – meaning it runs mainly on a single processor core – spend your money on the fastest clock speed processor with the most ample cache you can afford. Using this logic, a quad-core desktop processor like the Intel® Xeon® E5-1650v3 (at 3.8 GHz Max Turbo Frequency with 15 MB cache) or Intel® Xeon® E3-1281v3 (at 4.1 GHz Max Turbo Frequency with 8 MB cache) or Intel® Core™ i7-4790 (at 4.0 GHz Max Turbo frequency with 8 MB cache ) or quad-core mobile processor like the Intel® 4th Generation Core™ i7-4790MX (at 4.0 GHz Max Turbo Frequency with 8 MB cache) are a great choices for many SOLIDWORKS users. Of course in cases where the software is run concurrently with analysis and rendering software, it can make sense to move to more than four cores, but you’ll still want the highest speed and cache size you can afford.

Note: Intel Xeon 12xx/16xx series processors support more cache than the slightly less expensive i7 or i5 processors as well as supporting much more reliable error correcting code (ECC) memory. The additional reliability and speed are worth the extra cost.


So, if you are running other apps along with solidworks then then the extra threads and lower clock speed may be a good trade off. If you are just running solidworks then you will want to keep a system that has a higher clock speed as it will work better and be a better gaming rig, too.

Your i5 will likely be a much better system for some solidworks duties and regular gaming with it's high clock speed versus a much slower multi core or multiprocessor system. Maybe spend the upgrade budget on an i7 for your current MB? Better graphics for gaming or put that SSD RAID in there?

I have an EVGA SR-2 system (2P) I use daily and it is still worth bragging about as the only thing I've had to upgrade has been graphics and storage in several years. But, this was a very unique platform (Xeon overclocking) and I don't do professional work on this system. It's purely for my amusement.
 
Intel sold some Xeons unlocked (see my sig)...sadly never the high core ones. AMD could get a lot of sales if they unlocked them.

and no it isn't a bug...they (Intel) just don't unlock them because of profits. Doesn't mean AMD can't or wont

If they sold a 12-16 core chip unlocked i would sell my 1650v3 in a heart beat and buy that 16 core.

I don't see why AMD wouldn't sell an unlocked 16 core when Intel has a 10 core. AMD could get mad amount of business.


Wtf are you talking about multi SoC/ socket system? Linus used a high core CPU not mutli socket system. How about you watch video before talking out of your ass.

are you talking about this?
http://www.anandtech.com/show/10158/the-intel-xeon-e5-v4-review/2

vs this

http://www.anandtech.com/show/10337...x-6900k-6850k-and-6800k-tested-up-to-10-cores

I personally dont know anything about the delay moving from set to set in a CPU but internally in a CPU vs dual socket system is vastly different. Or you simply dont understand that AMD uses MCM for anything above 8 cores on a package. the outright horrible scaling of the 32 Naples chips is exactly due to that. 64 cores means 8 dies.

If you throw some 200 cycle bullshit number up back it up

I have never heard of thus penalty on xeons.

Unlocked Xeons isn't something people that buys them one. If you worked in the industry you know and you know why. Data integrity is everything. Including safe guarding it. Being unlocked is seen as a bug by the buyers.

Not sure why you link same die benchmarks. Obviously you didn't understand the issue and the issue presented with Windows and its like to throw threads around. Server applications isn't really affected by it due to their entirely different nature.

http://www.gdcvault.com/play/1022186/Parallelizing-the-Naughty-Dog-Engine
http://www.dualshockers.com/2014/03...l-and-how-they-can-make-them-run-really-fast/

It takes roughly 190 CPU cycles to access data in another core cluster's L2 cache, vs. 26 cycles to access data in a local L2 cache. At 1.6GHz, a Jaguar cycle is 0.625ns long. 190 cycles, therefore, is 118.75ns.

Now you learned something new today :)

NDPS4_8-571x425.jpg

NDPS4_7-568x425.jpg
 
Last edited:
Unlocked Xeons isn't something people that buys them one. If you worked in the industry you know and you know why. Data integrity is everything. Including safe guarding it. Being unlocked is seen as a bug by the buyers.

Not sure why you link same die benchmarks. Obviously you didn't understand the issue and the issue presented with Windows and its like to throw threads around. Server applications isn't really affected by it due to their entirely different nature.

http://www.gdcvault.com/play/1022186/Parallelizing-the-Naughty-Dog-Engine
http://www.dualshockers.com/2014/03...l-and-how-they-can-make-them-run-really-fast/



Now you learned something new today :)

NDPS4_8-571x425.jpg

NDPS4_7-568x425.jpg
I dont give a shit about the PS4. You made a statement that it is the same with Intel and AMD (Zen) MCM....prove it.

We all know AMDs old MCM are total trash....

I linked to 2 different CPU designs a small chip and medium chip. medium chips have several CPUs and small chip have a single CPU....i forget the term used. Small core count, Medium core count? SCC MCC design?
 
I dont give a shit about the PS4. You made a statement that it is the same with Intel and AMD (Zen) MCM....prove it.

We all know AMDs old MCM are total trash....

I linked to 2 different CPU designs a small chip and medium chip. medium chips have several CPUs and small chip have a single CPU....i forget the term used. Small core count, Medium core count? SCC MCC design?

You think there is no access penalty between dies and sockets?

You linked 2 things that is still the same die and still the same bus speed and cache. the PS4/Xbox One SoC is just the extreme of a same single die concept. Since its bolted together in the most primitive way.

MemoryHeirarchy.png
 
Last edited:
You think there is no access penalty between dies and sockets?

You linked 2 things that is still the same die and still the same bus speed and cache. the PS4/Xbox One SoC is just the extreme of a same single die concept. Since its bolted together in the most primitive way.

MemoryHeirarchy.png


so if intels 22 core Xeon is all the same die than your point is moot.....Intel doesnt have that issue....so what the fuck are you going on about if Intel is all the same die with none of these issues.

You keep throwing around SOC, die, and multisocket and changing all over the place when i said linus used an intel xeon to do this. If That is all the same die than wtf are you going on about. If AMD entripse chip isn't all on same die than that sucks but oh well.

Also i love how you say it is a bug to be able to overclock....it makes you sound ridiculous and gives me a good chuckle. Intel does sell unlocked ones too FYI but they are special order and probably a super premo over priced. (lutjens has talked about them but they are only sold to select major corps)

Again Intel doesnt sell unlocked because of money not because it would be a bug roflcopter.
 
Last edited:
I am talking about MCM designs and multi sockets systems and their penalties. The console SoC was just an example of one with the same penalties, despite being the same die. Zen server parts with 16 and 32 cores consist of 2 and 4 dies using MCM on a package with GMI interconnect. Same interconnect used across sockets.

You should check who the target segment is for Xeons. Seems you mixed it up. :)
 
I am talking about MCM designs and multi sockets systems and their penalties. The console SoC was just an example of one with the same penalties, despite being the same die. Zen server parts with 16 and 32 cores consist of 2 and 4 dies using MCM on a package with GMI interconnect. Same interconnect used across sockets.

You should check who the target segment is for Xeons. Seems you mixed it up. :)

lutjens wirk
again intel only locks them for money reasons hence why some very rich firms can buy unlocked ones.....this is well known now. Several hard members that have connections with intel/suppliers have gone over this in the past. Lutjens and wirk?

Intel usually fucks up and leaks the wrong slides with those chips but quickly white washes once they notice.

Again i stated in the beginning that linus used a xeon for this multi GPU multi gaming rig and you went on some tangent and said it would never work and included xeons in the mix for some odd reason.
 
Last edited:
Solidworks states their application is linear and relies heavily on single threaded performance for most workloads. Look at the system requirements for your apps - this is from a research paper on the solidworks site and a blog site about solidworks:


Processor

Your workstation’s engine is its processor and you only have one chance to get it specified properly. Since the main SOLIDWORKS application is “lightly threaded” – meaning it runs mainly on a single processor core – spend your money on the fastest clock speed processor with the most ample cache you can afford. Using this logic, a quad-core desktop processor like the Intel® Xeon® E5-1650v3 (at 3.8 GHz Max Turbo Frequency with 15 MB cache) or Intel® Xeon® E3-1281v3 (at 4.1 GHz Max Turbo Frequency with 8 MB cache) or Intel® Core™ i7-4790 (at 4.0 GHz Max Turbo frequency with 8 MB cache ) or quad-core mobile processor like the Intel® 4th Generation Core™ i7-4790MX (at 4.0 GHz Max Turbo Frequency with 8 MB cache) are a great choices for many SOLIDWORKS users. Of course in cases where the software is run concurrently with analysis and rendering software, it can make sense to move to more than four cores, but you’ll still want the highest speed and cache size you can afford.

Note: Intel Xeon 12xx/16xx series processors support more cache than the slightly less expensive i7 or i5 processors as well as supporting much more reliable error correcting code (ECC) memory. The additional reliability and speed are worth the extra cost.


So, if you are running other apps along with solidworks then then the extra threads and lower clock speed may be a good trade off. If you are just running solidworks then you will want to keep a system that has a higher clock speed as it will work better and be a better gaming rig, too.

Your i5 will likely be a much better system for some solidworks duties and regular gaming with it's high clock speed versus a much slower multi core or multiprocessor system. Maybe spend the upgrade budget on an i7 for your current MB? Better graphics for gaming or put that SSD RAID in there?

I have an EVGA SR-2 system (2P) I use daily and it is still worth bragging about as the only thing I've had to upgrade has been graphics and storage in several years. But, this was a very unique platform (Xeon overclocking) and I don't do professional work on this system. It's purely for my amusement.

When I said solidworks, I meant rendering solidworks animations, which is very multithreaded. Solidworks itself can be ran on an ultrabook, but the rendering will take forever if you don't have a fast CPU with lots of threads.
 
Back
Top