Evil Avatar  



Go Back   Evil Avatar > Daily Gaming News > News Items

» Sponsored Links


» Recent Threads
Top 10 Games EVERYONE...
Last post by vallor
Today 05:45 AM
7 Replies, 493 Views
Post something good you...
Last post by RAV
Yesterday 11:35 PM
411 Replies, 111,073 Views
Liberals gone wild
Last post by vallor
Yesterday 09:48 PM
6,849 Replies, 1,510,904 Views
Picts - Waiting to...
Last post by BeardedSonOfNel
Yesterday 09:22 PM
19,245 Replies, 4,253,532 Views
Kung flu
Last post by stalazon
Yesterday 08:18 PM
1,449 Replies, 60,086 Views
Weekend Headbanger -...
Last post by BeardedSonOfNel
Yesterday 06:29 PM
14 Replies, 525 Views
Weekend Gamer: What are...
Last post by BeardedSonOfNel
Yesterday 06:27 PM
10 Replies, 427 Views
Weekend Couch Potato -...
Last post by BeardedSonOfNel
Yesterday 06:22 PM
4 Replies, 225 Views
Reply
 
Thread Tools
Old 03-04-2017, 09:57 PM   #61
Jotoco
Evil Dead
 
Join Date: Aug 2006
Location: Campo Grande, MS - Brazil
Posts: 1,772
Quote:
Originally Posted by Anenome View Post
So let me get this right, our consensus then is that the best gamer chip is the 7700?
Since their price fell 25%, yes. Even in Brazil it is now.
Jotoco is offline   Reply With Quote
Old 03-04-2017, 10:23 PM   #62
Anenome
Autarchist
 
Anenome's Avatar
 
Join Date: May 2007
Location: Recursion City
Posts: 49,889
Blog Entries: 62
Intel's Cannonlake next, but when?

Intel's Cannonlake Leaked In A Patent

Intel: Cannonlake CPUs will be more than 15 percent faster than Kaby Lake

Intel: 10nm Cannon Lake PCs will start shipping in 2017

Hm, might be nice to wait for Cannonlake.

But wait...

Intel Corp. Delays Its 10-Nanometer Cannon Lake Chip - Again

Intel's upcoming 10nm-based Cannonlake will combat Ryzen
__________________
Choose your government: the majority ruling the minority, the minority ruling the majority, or everyone ruling themselves long as they do not initiate force, fraud, or theft against one another.
Anenome is offline   Reply With Quote
Old 03-04-2017, 11:09 PM   #63
SpectralThundr
Evil Dead
 
SpectralThundr's Avatar
 
Join Date: Oct 2013
Location: Boston/Ontario
Posts: 9,843
Quote:
Originally Posted by Anenome View Post
So let me get this right, our consensus then is that the best gamer chip is the 7700?
Well it's certainly not an amd chip. Apparently they just don't have the resources or talent to actually compete.
SpectralThundr is offline   Reply With Quote
Old 03-05-2017, 12:42 AM   #64
BenSkywlkr
Evil Dead
 
Join Date: Mar 2013
Posts: 403
Quote:
Ben, you appear out of your league here with virtualization. It's not about individual core performance, it's about the number of cores.
Lol, wow kid, OK, we can go there.

What type of impact do you think the cache structure and the implicit sacrifices made in vector ops is going to have on VM performance given AMD's explicit statement that their HT implementation is broken on a core level working under Windows combined with the lack of quad channel RAM when trying to virtualize out and having to constantly flush buffers? Furthermore, how do you propose to overcome these issues at the core OS level and what kind of implementation on a software level do you see MS being able to implement outside of a rigid cache allocation system which as you must be an expert on, can lead to outright system crashes when attempting to use such techniques for a modern OS VM setup?

To those general points, how do you propose to effectively utilize VMs on a platform that supports a minuscule 4GB of RAM per virtual core, and to compound that, only in a paltry dual channel configuration with low clock speeds and *very* high latency?

On a continuation of your claimed staggering expertise on the subject, how are you going to structure your code base to perform tasks in a virtual machine faster then they do natively as is required to make your claims valid.

Quote:
1) This is a consumer-level chip.
That is apparent in absolutely everything about the CPU, except performance for things consumers would use them for.

Quote:
2) More cores means more VM's have a dedicated core as opposed to sharing. This is huge.
No- 'this is huge' is an absolute- that statement is only valid for a minuscule subset of instances. If all of your cores and schedulers are taxed to their maximum capacity then having more cores *CAN* be of benefit- if you aren't thrashing data in and out of caches and overrunning your RAM allocation at which point you would be *MUCH* better served to simply hold the VM that will cause the stall until you have idle time. You are correct that less cores wouldn't help you here, but less cores that were markedly faster would be superior.

Quote:
3) With more cores I can spin up more VM's
Of what? Win98? You don't have enough cache, you don't have enough RAM to do this kind of work on these chips.

Quote:
4) With that in mind, buying into an 8 core from AMD is cheaper, MUCH cheaper.
This is very true- *EXACTLY* like it was with Bulldozer and Piledriver. If you think MOAR COORS! is the answer, AMD is without a doubt the company for you.
BenSkywlkr is offline   Reply With Quote
Old 03-05-2017, 08:17 AM   #65
brandonjclark
Evil Dead
 
brandonjclark's Avatar
 
Join Date: Dec 2007
Posts: 11,177
Quote:
Originally Posted by BenSkywlkr View Post
Lol, wow kid, OK, we can go there.
I'm 38 years old. Treat me with some respect.

Quote:
Originally Posted by BenSkywlkr View Post
What type of impact do you think the cache structure and the implicit sacrifices made in vector ops is going to have on VM performance given AMD's explicit statement that their HT implementation is broken on a core level working under Windows combined with the lack of quad channel RAM when trying to virtualize out and having to constantly flush buffers? Furthermore, how do you propose to overcome these issues at the core OS level and what kind of implementation on a software level do you see MS being able to implement outside of a rigid cache allocation system which as you must be an expert on, can lead to outright system crashes when attempting to use such techniques for a modern OS VM setup?
Who cares about HT? I told you this was about cores. If I bought this chip to use as my primary gaming PC and then down the road threw it in a headless server (as my current Opteron is), I'm going to disable HT.

Quote:
Originally Posted by BenSkywlkr View Post
To those general points, how do you propose to effectively utilize VMs on a platform that supports a minuscule 4GB of RAM per virtual core, and to compound that, only in a paltry dual channel configuration with low clock speeds and *very* high latency?

Here you go again talking like non-mission critical workloads must be all about performance for the HOME consumer. I don't know what you mean by "effectively utilize". Are you asking if I'd be able to implement my VM's to their max performance without running into CPU scheduling errors on the hypervisor level? What the hell do you think I'm running at home? These aren't terribly important workloads. I've got a file server, a Citrix implementation which takes up 5 VM's, a minecraft server, a Puppet server, a Plex server, and finally a torrent box.

Quote:
Originally Posted by BenSkywlkr View Post
On a continuation of your claimed staggering expertise on the subject, how are you going to structure your code base to perform tasks in a virtual machine faster then they do natively as is required to make your claims valid.
Are you just googling these things? These aren't even good questions. And who told you I was writing code to take advantage of the Ryzen virtualization platform? Are you making this up? Whether you understand this or not Ben, people virtualize at HOME for many reasons, and one of my reasons is I like to work with the tech. Another reason is I like to separate my boxes. Why would I load ALL of those applications I listed earlier onto just one un-virtualization OS install? What do you think this is, the 90's?


Quote:
Originally Posted by BenSkywlkr View Post
That is apparent in absolutely everything about the CPU, except performance for things consumers would use them for.
Okay...

Quote:
Originally Posted by BenSkywlkr View Post
No- 'this is huge' is an absolute- that statement is only valid for a minuscule subset of instances. If all of your cores and schedulers are taxed to their maximum capacity then having more cores *CAN* be of benefit- if you aren't thrashing data in and out of caches and overrunning your RAM allocation at which point you would be *MUCH* better served to simply hold the VM that will cause the stall until you have idle time. You are correct that less cores wouldn't help you here, but less cores that were markedly faster would be superior.
Do you have many DRS rules setup at home? Great, most people don't and I don't think we need to bring this up here. But since, you did....

What matters most at home? Efficiency, right? If I have an Intel chip running at max load on all cores while on the CUSP of ballooning memory (peak efficiency) what is my power usage compared to the Ryzen chip? If I allow the hypervisor to auto-schedule, which I should, I'm going to see MUCH less power draw. In your case, if I was running mission critical workloads or, hell, even something like mining, I suppose I would care about max performance. I don't. Not at home. I care about efficiency and a large portion of that equation for me are power costs. Again, at home.

Quote:
Originally Posted by BenSkywlkr View Post
Of what? Win98? You don't have enough cache, you don't have enough RAM to do this kind of work on these chips.
Hmm, I'm starting to wonder if you've ever virtualized anything. At all. Ever.





See that? That's on an Opteron system with 4 cores and 32 GB of RAM. Now, if I wanted I could spin up many more VM's, OVERPROVISIONING my RAM, CPU and HDD if I wanted. And guess what? There would be NO problem with that, until I tried to maximize performance across enough VM's to encounter thrashing. Modern datacenters depend on this over-provisioning my friend. They DEPEND on it. There are in fact people who sit and calculate capacity all the time with this principle in mind. How do you think the cloud works for AWS or Azure? Do you think they have one cpu for each virtual cpu? Now, if I was put to task to architect an infrastructure across a few datacenters then power would be considered cheap in that scenario compared to the compute power I wanted to maximize. At home, I only need to worry about max capacity initially, and then power costs come into play the rest of the time.

Finally, I understand your point. If my goal was to maximize cycles regardless of power, VM count and memory limits, I MIGHT want an Intel CPU at home. But as of right now, I'd rather have more cores running at lower draw. A minecraft server, Ben. I'm not folding proteins.

Quote:
Originally Posted by BenSkywlkr View Post
This is very true- *EXACTLY* like it was with Bulldozer and Piledriver. If you think MOAR COORS! is the answer, AMD is without a doubt the company for you.
More cores ARE the answer to my home server virtualization needs. I'm sorry you're unable to see that.

BTW, are you going to address the link someone put up earlier? Those look like some pretty amazing benchmarks to me.

https://www.servethehome.com/amd-ryz...ux-benchmarks/

from that link

Quote:
Here is the cool part for those programmers reading, 8 core/ 16 thread Ryzen 7 1700X is equivalent to $730-$770/ month of AWS compute using the c4.4xlarge and c4.8xlarge instances as guideposts. Both offer RAM capacities that Ryzen 7 1700X systems can handle.
__________________
~B$
Gamertag: legisilverback | Steam Nickname: brandonjclark
...playing Rebel Galaxy

Last edited by brandonjclark; 03-05-2017 at 08:54 AM..
brandonjclark is offline   Reply With Quote
Old 03-05-2017, 06:12 PM   #66
BenSkywlkr
Evil Dead
 
Join Date: Mar 2013
Posts: 403
Quote:
I'm 38 years old. Treat me with some respect.
I was writing code while you were still shitting your pants. I will return the level of respect I am shown.

Quote:
I told you this was about cores.
You also told me this was going to be a better gaming chip then an i7.

Quote:
Here you go again talking like non-mission critical workloads must be all about performance for the HOME consumer.
We are discussing $500 CPUs, if you don't care about performance as a main priority and are spending this kind of money quite frankly you are fucking retarded.

Quote:
Why would I load ALL of those applications I listed earlier onto just one un-virtualization OS install? What do you think this is, the 90's?
You are showing your youth and inexperience here kid

We started off with all of these things being virtualized off of Mainframes at least we were in the 70s and 80s, they tried to bring that back in the 90s(see Sun) and it failed because they couldn't hope to compete with local. Right now virtualization is seeing a resurgence due to people wanting desktop class operations in mobile form factors. Virtualizing a desktop environment on a desktop reduces performance, increases workload, reduces efficiency, pretty much there is no reason to do it unless you are just playing around with the tech(which is entirely valid in and of itself).

Quote:
What matters most at home? Efficiency, right?
Do you think I'm an inbred monkey kicked in the head repeatedly by a donkey? What kind of half witted moron uses x86 anything that cares about efficiency...? That is like buying a full size pick up truck to use as a canoe- what are you even saying?

Quote:
If I have an Intel chip running at max load on all cores while on the CUSP of ballooning memory (peak efficiency) what is my power usage compared to the Ryzen chip?
Latest and greatest AMD wet dream chip against four year old ARM, it's a slaughter. Let's not be absolute morons here- custom Linux ARM build just beats anything x86 stupid on every level of efficiency by huge margins, this isn't a contest, bring up a valid point please.

Quote:
Modern datacenters depend on this over-provisioning my friend.
That was the enitrety of MY point. HAF- Hurry up and finish- less cores that are significantly faster will outperform more physical cores *UNLESS* you are frequently maxing out system resources. It sounds like you are starting to wrap your head around the general idea of why these will suck compared to comparable Intel chips at least.

Quote:
More cores ARE the answer to my home server virtualization needs.
If that makes you sleep better at night, some of my kids like fairy tales before they sleep too.

Quote:
BTW, are you going to address the link someone put up earlier? Those look like some pretty amazing benchmarks to me.
/facepalm

I have, explicitly. I have stated that Ryzen is a beast at GPGPU benchmarks as long as you ignore the existence of GPUs.
BenSkywlkr is offline   Reply With Quote
Old 03-05-2017, 09:42 PM   #67
Rommel
International Playboy
 
Rommel's Avatar
 
Join Date: Feb 2005
Location: Hong Kong
Posts: 10,325
I frankly do not know why you two are arguing over the actual make of the CPU, I see it as irrelevant. As a consumer rather than one of its developers, the first point that should be considered is performance. And when I check the benchmarks for performance, Ryzen loses and it loses every time. There are a few use cases, valid or otherwise, where it competes and does so with more expensive units to boot. It still loses. There are a number of use cases, ones I see as more common, where it cannot compete with components half its price. That's a big lose.

After you have examined those benchmarks and the bevy of losses, I ask whether we should bother diving into the nitty gritty of its usage for virtualizing four desktops rather than giving each cubicle its own machine. Does it really matter that those use cases exist when we already know who wins in those scenarios?

Especially when no reviewer can get the God-damn RAM to work! When a car comes off the line defective, you don't debate how nice is its paint job.
__________________
Quotes in context.*
Quote:
Originally Posted by Rommel View Post
drive into the daughter's womb proper, fill the uterus with the dick snot
Quote:
Originally Posted by screwyluie View Post
Rommel gave the right answer
*Not in context
Rommel is offline   Reply With Quote
Old 03-06-2017, 09:17 PM   #68
Clancy
Soul Reaper
 
Join Date: Feb 2005
Posts: 223
Quote:
Originally Posted by Rommel View Post
I frankly do not know why you two are arguing over the actual make of the CPU, I see it as irrelevant. As a consumer rather than one of its developers, the first point that should be considered is performance. And when I check the benchmarks for performance, Ryzen loses and it loses every time. There are a few use cases, valid or otherwise, where it competes and does so with more expensive units to boot. It still loses. There are a number of use cases, ones I see as more common, where it cannot compete with components half its price. That's a big lose.
There is only one segment where Ryzen doesn't perform as well as it should, and that is gaming at 1080p, and the reason for this was already addressed, and it also has higher minimum FPS as well compared to 7700K.
This is going to be addressed by Microsoft when they change the Kernel to support Ryzen.
In 1440p or 4K, Ryzen doesn't lose many benchmarks at all.
The cheapest Ryzen out so far is $330, and that does bring value once you take off your rosey colored glasses and actually look what is going on, and what you can do with it.

https://www.youtube.com/watch?v=8-mMBbWHrwM

Quote:
Especially when no reviewer can get the God-damn RAM to work! When a car comes off the line defective, you don't debate how nice is its paint job.
Just like the X99's RAM issues when it first came out?

"Intel X99 Motherboard Goes Up in Smoke For Reasons Unknown" http://www.legitreviews.com/intel-x9...unknown_150008

"My Core i7 5960X + MSI X99 Experience So Far: It's Smoking, Really" http://www.phoronix.com/scan.php?pag...x99_fail&num=1

Quote:
As a result, the motherboard manufacturers were not as prepared as usual to qualify all the memory ready to go on sale (this has been fixed now with the latest BIOS updates) and memory manufacturers are still putting kits together. Combined with the high price of DDR4 and the initial limited quantity, prices for DDR4 were particularly high: $250 for 16GB as we noted at Haswell-E launch.
Memory manufacturers are still preparing kits, and motherboard manufacturers are continuously updating their qualified lists. With all the memory kits I have in to test, I have not had a motherboard not POST because of it, however in order to get the best performance out of that new high end memory kit (2666 MHz+), BIOS updates should be the number one priority. Most manufacturers have software that can do this (MSIís Live Update 6) or users can download and update manually.
http://www.anandtech.com/show/8557/x...ntel-haswell-e


On and on it went... it is as if y'all have selective amnesia.
New platforms pretty much ALWAYS have problems. They will fix Ryzen issues, just like they have (mostly) fixed the X99 issues.

You guys are just drama queens.
Clancy is offline   Reply With Quote
Old 03-07-2017, 06:43 AM   #69
Anemone
Evil Dead
 
Anemone's Avatar
 
Join Date: Mar 2013
Location: Autarchist
Posts: 6,092
Blog Entries: 3
Thanks, Clancy. Voice of reason.
__________________
Member of the Nintendo Offensive Front.
Anemone is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 06:13 AM.