Time to build a new rig ???

[Crystallas]

Hes probably gonna use it for everything which isn't a good idea.... you don't want to use an SSD for regular storage duties.... put your OS and your games and programs that will demand alot of disk IOPS if you are using the machine for something else like maybe heavy video editing or SQL Server or something.

Have a regular cheaper drive or set of drives in RAID where you store junk like videos, pictures, documents, porn, etc.... 20$ says the kid has like 500GB of porn hes gonna store on a fucking SSD :lmao:

:yep:

My main rig has 6 disks. 4 SSDs and 2 HDDs, and I sure as hell don't use SSDs for any long term storage.

2x 32gb SSDs R0 striped. One partition is 24gb, the other partition is ~32gb and assigned for caching duties of various programs. Because windows doesn't care about swap partitions, I just use the cache area as the allocated virtual memory. 24+32 =/= 64.. yes, that's formatted capacity, kiddies.

A 250gb OS drive, this has my Arch installation.

A 240gb Win/VM drive. I have it partitioned in 3 spots. Most for windows 8.1. A partition with virtual machines images for Win 7, although I just VM Win8.1 now that I got it just right to VM and boot without issues.

My ~user drive, which has games, downloads to be filtered, working projects, etc. 2tb.

And my secure drive, which is stacked with encryption. Has all my business and financial data, other projects that I'm not currently working on. Stuff I need to have immediate access to for various reasons, but use rarely.

Then we get into the home server, which also acts as a secure LAN/NAS. That's a whole different story, LOL. But I guess I use a SSD boot drive in that, Compactflash with a blazing fast ATA133 connection


The only place where anyone would benefit (both long and short term) with a SSD larger than 100gb, is if the price was within reason, your system acts as a quick seek file host... and that's it. Anything else, you're better off doubling your memory.

 

[botfly10]

wat crys said about intel CPU's.

Also, first consideration is your budget. You gotta know that cause it affects every configuration choice you make. If you are capped at $900 vs $1200, it will change your approach a lot.

 

[botfly10]

I have this CPU and mobo and I love the combo. Fucking crushes

Intel 4770K
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116901

Asus maximus IV hero mobo
http://www.newegg.com/Product/Produ...gclid=CJSs7e3dr8QCFQmVfgod6CMAAg&gclsrc=aw.ds

Those components are a little old now, but just for reference, I don't think anything has ever bottle-necked on my cpu.

This is probably the current equivalent to what I have:

i7-4790K
http://www.newegg.com/Product/Product.aspx?Item=N82E16819117369

 

[AussieBear]

:yep:

My main rig has 6 disks. 4 SSDs and 2 HDDs, and I sure as hell don't use SSDs for any long term storage.

2x 32gb SSDs R0 striped. One partition is 24gb, the other partition is ~32gb and assigned for caching duties of various programs. Because windows doesn't care about swap partitions, I just use the cache area as the allocated virtual memory. 24+32 =/= 64.. yes, that's formatted capacity, kiddies.

A 250gb OS drive, this has my Arch installation.

A 240gb Win/VM drive. I have it partitioned in 3 spots. Most for windows 8.1. A partition with virtual machines images for Win 7, although I just VM Win8.1 now that I got it just right to VM and boot without issues.

My ~user drive, which has games, downloads to be filtered, working projects, etc. 2tb.

And my secure drive, which is stacked with encryption. Has all my business and financial data, other projects that I'm not currently working on. Stuff I need to have immediate access to for various reasons, but use rarely.

Then we get into the home server, which also acts as a secure LAN/NAS. That's a whole different story, LOL. But I guess I use a SSD boot drive in that, Compactflash with a blazing fast ATA133 connection


The only place where anyone would benefit (both long and short term) with a SSD larger than 100gb, is if the price was within reason, your system acts as a quick seek file host... and that's it. Anything else, you're better off doubling your memory.

what do you do? hosting? reseller? ???

 

[AussieBear]

wat crys said about intel CPU's.

Also, first consideration is your budget. You gotta know that cause it affects every configuration choice you make. If you are capped at $900 vs $1200, it will change your approach a lot.

the budget can fluctuate depending on if i want extra monitors and the such. im leaning towards amd, but now im going to wait a 3-4 months until after i buy a new car, because if i spend less there, it'll go towards the rig.

 

[Crystallas]

the budget can fluctuate depending on if i want extra monitors and the such. im leaning towards amd, but now im going to wait a 3-4 months until after i buy a new car, because if i spend less there, it'll go towards the rig.

Good plan.

I've been telling people to wait if they can. In over 30 tears of computer building, I have only told people to wait twice. Except I wasn't as adamant about telling you to wait, because those who do not know my track record of recommending current solutions otherwise will start debating as if I take an always wait approach. :facepalm: But right now, the more you can wait, the better. Too many new standards have rolled out, and some are not even fully implemented. So either you can jump on next gen for similar prices that you would pay now, or jump on prior gen for much cheaper. Unusual. I've explained this phenomenon before in another post, and how it has only happened twice in x86 before this time.

Oh, and I'm an engineer of many hats. Mainly electrical and audio. Thus why I have a goofy system with different needs.

 

[AussieBear]

I have this CPU and mobo and I love the combo. Fucking crushes

Intel 4770K
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116901

Asus maximus IV hero mobo
http://www.newegg.com/Product/Produ...gclid=CJSs7e3dr8QCFQmVfgod6CMAAg&gclsrc=aw.ds

Those components are a little old now, but just for reference, I don't think anything has ever bottle-necked on my cpu.

This is probably the current equivalent to what I have:

i7-4790K
http://www.newegg.com/Product/Product.aspx?Item=N82E16819117369

i wouldnt mind that 4790 but since the aussie dollar is shit im paying newegg roughly 450 not including shipping. i have that, im just a price to performance dinosaur.

locally i could have the amd fx 9590 with a liquid cooler for 275 bucks.

im not sure what kind of board im going with. i need to decide if i want to build a mini or full tower.

 

[botfly10]

Crys, in your estimation what is the time threshold for when this shit will start being implemented and supported by software? Implemented on the consumer level and actually utilized by applications. Months? Years?

 

[Crystallas]

Crys, in your estimation what is the time threshold for when this shit will start being implemented and supported by software? Implemented on the consumer level and actually utilized by applications. Months? Years?

What do you mean? Once something is out, it's supported nearly instantly in the major platforms. For example, if you read the lists for linux kernel development, for x11 development, betawires for games, a lot of features for future hardware gets built in when possible. It's amazing in some examples, like broadwell was supported almost a year in the linux kernel, before the CPU was released. The R9 300 Radeons are already natively supported in a 2 month old kernel. GL rendering is supported in software sometimes before the hardware is out. PCIe 4, SATA 3.2, USB 3.1, Displayport 1.3, FreeSync, Skylake, were all implemented and supported under linux within days of specification release and months before the actual hardware releases. OSX/iOS benefits from some of this development, as it basically rips off GNU code and closes itself, thus why Apple supports more products now, than it has in the past. Also why linux changed how the kernel is structured, because they're sick of apple claiming to innovate something with their cherry picked developers, only to be a near bit-matched clone of someone else's software. I only mention this part, because OSX will likely take the longest to implement hardware features yet again.

For windows, yes, I use windows too. It's a completely different hierarchy. If the hardware has MS's blessing, it will be supported out of the box. What is MS's blessing? Basically share your proprietary code with them, as quality checks aren't really done on hardware. But like any OS, you need to jump through hoops to use drivers that aren't open enough to be natively added into the initialization stage of the OS. Since MS has switched to git for development, things should get better in areas that were lacking before.

TL;DR
Some things are supported on Day-0. Some things take weeks. It's rare when a new feature takes months to be implemented(and often that means it's an obscure feature with limited demand to begin with). The big difference is within totally new standards and new versions of existing standards.

Right now, the only two standards are going through maturing phases in the mainstream market. Video buses with the 4k+ signal standards(which is why I don't even bother suggesting anything less than displayport 1.3, because dp1.2, HDMI 2.0 both have a number of issues that might or might not show up and are hard to troubleshoot). The other is DDR4. Don't let the number fool you, it's not an existing standard that simply has a newer version, it's a wildly different standard and not mature. The existing memory controllers do not take advantage of it, and that's not a software limit, but a hardware limit, as well as part of the bus problem we're running into.

 

[AussieBear]

What do you mean? Once something is out, it's supported nearly instantly in the major platforms. For example, if you read the lists for linux kernel development, for x11 development, betawires for games, a lot of features for future hardware gets built in when possible. It's amazing in some examples, like broadwell was supported almost a year in the linux kernel, before the CPU was released. The R9 300 Radeons are already natively supported in a 2 month old kernel. GL rendering is supported in software sometimes before the hardware is out. PCIe 4, SATA 3.2, USB 3.1, Displayport 1.3, FreeSync, Skylake, were all implemented and supported under linux within days of specification release and months before the actual hardware releases. OSX/iOS benefits from some of this development, as it basically rips off GNU code and closes itself, thus why Apple supports more products now, than it has in the past. Also why linux changed how the kernel is structured, because they're sick of apple claiming to innovate something with their cherry picked developers, only to be a near bit-matched clone of someone else's software. I only mention this part, because OSX will likely take the longest to implement hardware features yet again.

For windows, yes, I use windows too. It's a completely different hierarchy. If the hardware has MS's blessing, it will be supported out of the box. What is MS's blessing? Basically share your proprietary code with them, as quality checks aren't really done on hardware. But like any OS, you need to jump through hoops to use drivers that aren't open enough to be natively added into the initialization stage of the OS. Since MS has switched to git for development, things should get better in areas that were lacking before.

TL;DR
Some things are supported on Day-0. Some things take weeks. It's rare when a new feature takes months to be implemented(and often that means it's an obscure feature with limited demand to begin with). The big difference is within totally new standards and new versions of existing standards.

Right now, the only two standards are going through maturing phases in the mainstream market. Video buses with the 4k+ signal standards(which is why I don't even bother suggesting anything less than displayport 1.3, because dp1.2, HDMI 2.0 both have a number of issues that might or might not show up and are hard to troubleshoot). The other is DDR4. Don't let the number fool you, it's not an existing standard that simply has a newer version, it's a wildly different standard and not mature. The existing memory controllers do not take advantage of it, and that's not a software limit, but a hardware limit, as well as part of the bus problem we're running into.

yeah during my research ive only seen a limited amount of cpus (Haswell-E and the Wellsburg X99) that support ddr4 and even less mobos. currently, no market amd cpu has any with support but they said they will in the future.

 

[Crystallas]

DDR4 was designed to solve(more like improve on) 3 problems with DDR3.

power(which also translates to heat)
per module capacity and bandwidth
open row encoding

None of these solutions result in faster performance over DDR3 with existing or similar modules. When better stacking modules are used, then DDR4 will become much faster, namely because a stacked module decreases travel considerably, which also means the way CAS latency is measured will finally become obsolete as a metric to determine the difference in performance between similarly clocked sticks using the same or similar components.

Right now, even the Haswell DDR4 controller is poorly optimized. Intel has always been ahead with implementing new memory standards, but their memory controllers are often lacking for a few iterations. Then that sends shockwaves through the industry, as memory manufacturers try to engineer the best they can based on intel's controller. In the past, that was a battle between micron and corsair to figure out who does what better with most of the industry building on what these two companies released. But now that Samsung is a major player in memory, both micron and corsair are walking on shells to avoid lawsuits instead of push development. DDR4 is one standard that nobody should be in any hurry to buy into, unless it happens to be the memory controller on a board/cpu that make up for it. DDR3 will be the better, more practical, cheaper, and even faster standard for another 6 months or more.

 

[botfly10]

What do you mean? Once something is out, it's supported nearly instantly in the major platforms. For example, if you read the lists for linux kernel development, for x11 development, betawires for games, a lot of features for future hardware gets built in when possible. It's amazing in some examples, like broadwell was supported almost a year in the linux kernel, before the CPU was released. The R9 300 Radeons are already natively supported in a 2 month old kernel. GL rendering is supported in software sometimes before the hardware is out. PCIe 4, SATA 3.2, USB 3.1, Displayport 1.3, FreeSync, Skylake, were all implemented and supported under linux within days of specification release and months before the actual hardware releases. OSX/iOS benefits from some of this development, as it basically rips off GNU code and closes itself, thus why Apple supports more products now, than it has in the past. Also why linux changed how the kernel is structured, because they're sick of apple claiming to innovate something with their cherry picked developers, only to be a near bit-matched clone of someone else's software. I only mention this part, because OSX will likely take the longest to implement hardware features yet again.

For windows, yes, I use windows too. It's a completely different hierarchy. If the hardware has MS's blessing, it will be supported out of the box. What is MS's blessing? Basically share your proprietary code with them, as quality checks aren't really done on hardware. But like any OS, you need to jump through hoops to use drivers that aren't open enough to be natively added into the initialization stage of the OS. Since MS has switched to git for development, things should get better in areas that were lacking before.

TL;DR
Some things are supported on Day-0. Some things take weeks. It's rare when a new feature takes months to be implemented(and often that means it's an obscure feature with limited demand to begin with). The big difference is within totally new standards and new versions of existing standards.

Right now, the only two standards are going through maturing phases in the mainstream market. Video buses with the 4k+ signal standards(which is why I don't even bother suggesting anything less than displayport 1.3, because dp1.2, HDMI 2.0 both have a number of issues that might or might not show up and are hard to troubleshoot). The other is DDR4. Don't let the number fool you, it's not an existing standard that simply has a newer version, it's a wildly different standard and not mature. The existing memory controllers do not take advantage of it, and that's not a software limit, but a hardware limit, as well as part of the bus problem we're running into.

sigh

In your estimation, what is it that is worth waiting for and how long until it comes out and is being taken advantage of?

 

[botfly10]

You know what Einstein said about explaining things simply...

 

[Crystallas]

You know what Einstein said about explaining things simply...

I know, I should get a damn Nobel prize for simplifying something that is about a thousand times more complex.

sigh

In your estimation, what is it that is worth waiting for and how long until it comes out and is being taken advantage of?

Display interface standards are maybe the biggest one worth waiting for. All of the existing solutions are bandaid fixes to extend the life of soon to be dead interface standards. Or they simply aren't available to the consumer from start to finish of the i/o structure. But we're close. Quite a few new(recently released, like in the last few weeks) monitors are now supporting the next gen cables with full spec and standards(not just doubling up existing cables to work outside of their maximum spec), a frame lock/sync function. That's good news, and it's up to the user as far as how long they want to wait, because having more choice helps narrow down the decision. The video devices that support these standards are all next gen cards, as not a single one is out in single card. On top of that, the new VESA interface standards, like displayport 1.3 is backwards compatible with some of the current cards, like his HD 280, so it would just be about waiting for the freesync drivers to come out in his case, with a freesync capable monitor(I think 3 are purchasable now, with another 14 or so left to come out in the next few months). This standard is around the corner, weeks? Taken advantage of, as soon as the hardware is out. It's a firmware solution as far as games are concerned, so if they support the output, they don't care how it happens.

z170 is another, but who knows when that will come out. It's part of skylake, and the combo of z170 and skylake's architecture on the new LGA 1151 is a huge bundle of fixes, including reaping the benefits from DDR4 finally. Not to mention it's not as half-assed as broadwell. Even if he opted for the upcoming Toronto core, it's a huge real world improvement over the already dated Vishera systems he was looking at, and should be very cost effective.

 

[botfly10]

I appreciate your effort, I'm just a layman is all.

So, sounds like you are saying new monitor and gpu shit is gonna do better with frame syncing? And maybe better support of 4k resolution or something? Not sure what else the port and cable shit you mention does from a consumer experience perspective.

Sounds like new mobo architecture or some shit will use DDR4 to its potential. Whats that mean? More than 4gb ram per chip or whatever the current limit is? Whats that mean from a consumer experience perspective?

 

[Crystallas]

Layman explanation.

Time travel scenario. Let's say you own a Playstation 2 as your console, games were made for it until 2013! Do you buy the PS3 or wait 2 months for a PS4? That's the magnitude of the many display and GPU standards going into place not only for 4k, syncing, but also new directx(windows10). Or do you buy the PS3 now? Two philosophies, and of course, a computer is far more useful with a larger library of games and software, but it's comparable.

 

[botfly10]

Cool. Thats pretty damn exciting. I hear great things about the geforce gsync shit already

 

[Wintermute]

All good information gents. To summarize:

Totally agree about having more cores not equaling performance gains. Still running with a phenom IIx4 980 (3.7) and have no issue with running any game I play. Don't think it will be pried out of my cold, dead hands until the FX-8350 comes WAY down in price, AMD's next CPU does much better (Carrizo based or FX), or my phenom stops taking care of business (least likely)

Also, you can't go wrong with Samsun SSD. I have both the 830 256 GB in my tower, and bought the 840 EVO 120GB for my bargain laptop. Talk about a cheap way to squeeze some speed out of a cheap notebook! Here's a link to some good info on SSDs.

http://www.tomshardware.com/reviews/ssd-recommendation-benchmark,3269.html

The above site has really great info on other components too. I especially like their hierarchy charts for CPU and GPUs - really helps get a rough idea about whether or not one will see real-world performance gains when upgrading.

Got all excited about one of the last generation of Samsung monitors and they turned out to have major quality control issues. I think the story at the time (circa 08/09) was their LCD division was having big issues. No worries though, picked up a really fast Viewsonic monitor and still love it. 120 Hz refresh, and excellent for FPS.

Last and least, also have a recently purchased R9-280X Tri-X OC edition. Just love it. Made a nice jump in framerates from my old 6950 (that I flashed to 6970 shaders, bwahahaha!). So I think your card has got plenty of muscle.

Anyway good luck with your system.

 

[xer0h0ur]

If your aim is at using a modern AMD cpu then wait for the Zen next year otherwise just go with an X99 based cpu like a 5820K. For what its worth, DX12 in Windows 10 is going to level the playing field hardcore. Even for AMD's current multi-core offerings. This is because DX12 is designed to load balance across all CPU cores and its particularly going to be optimized for 6 core processors. The other major feature of DX12 is asynchronous shaders which will unleash the compute power in AMD's GPUs which was typically sitting around playing patty cake while gaming since nothing was utilizing it.

 

[Crystallas]

Not just DX12. But Vulkan is already showing some nice benefits.