|
When using this resource, please read the opening post. The Tech Support forum regulars have helped create countless of desktop systems without any compensation. The least you can do is provide all of the information required for them to help you properly. |
On November 20 2013 02:25 Disregard wrote:
I play games too, especially the Total War series. So I never intended to replace my ancient gaming PC build with workstation parts. But if its feasible for both applications I don't mind.
edit: So besides some extensions that I will never utilze and lower TDP, the Ivy Bridge 1240 is essentially the same as the Haswell 1240? Seems like the Xeons for me are a better deal overall compared to 4xxx series.
It works like this, using the Xeon E3-1230 as an example:
E3-1230 <--- Sandy Bridge internally, uses socket LGA1155
E3-1230 v2 <--- Ivy Bridge, uses socket LGA1155
E3-1230 v3 <--- that one's Haswell, uses socket LGA1150
You want the ones with "v3" in their name. Those are the ones with most current architecture that's used in the i5/i7-4xxx CPUs. They physically fit into the normal consumer LGA1150 boards, and the typical BIOS should support them.
+ Show Spoiler [something more about how the names work] +An E3-1235 v3 would have integrated graphics.
The E3-1230 v3 is the lowest you should look at. The 1220 is weird: it does not have HyperThreading, but it's also not an i5 as it still has the increased cache of the i7.
|
On November 20 2013 00:00 iTzSnypah wrote:
DDR4 is going to be crazy expensive and everyone is going to stay with DDR3 until prices become reasonable (just like when DDR3 came out, it wasn't fully adopted for over 2 years). My motherboard was MSI's flagship when DDR3 came out and it was their only motherboard to support DDR3 during that time. How many pins is DDR4 though? Still 240?
About 90$-100$ for 8GB of DDR4 when ready for consumer market, expected to drop to current DDR3 prices after 2 years. Is that a normal cycle? Technology transitions so much faster these days.
|
On November 20 2013 04:22 Ropid wrote:Show nested quote +On November 20 2013 02:25 Disregard wrote:
I play games too, especially the Total War series. So I never intended to replace my ancient gaming PC build with workstation parts. But if its feasible for both applications I don't mind.
edit: So besides some extensions that I will never utilze and lower TDP, the Ivy Bridge 1240 is essentially the same as the Haswell 1240? Seems like the Xeons for me are a better deal overall compared to 4xxx series. It works like this, using the Xeon E3-1230 as an example: E3-1230 <--- Sandy Bridge internally, uses socket LGA1155 E3-1230 v2 <--- Ivy Bridge, uses socket LGA1155 E3-1230 v3 <--- that one's Haswell, uses socket LGA1150 You want the ones with "v3" in their name. Those are the ones with most current architecture that's used in the i5/i7-4xxx CPUs. They physically fit into the normal consumer LGA1150 boards, and the typical BIOS should support them. + Show Spoiler [something more about how the names work] +An E3-1235 v3 would have integrated graphics.
The E3-1230 v3 is the lowest you should look at. The 1220 is weird: it does not have HyperThreading, but it's also not an i5 as it still has the increased cache of the i7.
Thanks, seems I was behind considering the mainstream market rarely goes for the Xeon and opts for the non-k when your typical gamer nowadays have dedicated GPUs.
edit: Seemingly my posts make me sound like a old man struggling to educate myself with modern technology...
|
The random read is pretty weak thouhgh.
|
On November 20 2013 05:09 Disregard wrote:Show nested quote +On November 20 2013 00:00 iTzSnypah wrote:
DDR4 is going to be crazy expensive and everyone is going to stay with DDR3 until prices become reasonable (just like when DDR3 came out, it wasn't fully adopted for over 2 years). My motherboard was MSI's flagship when DDR3 came out and it was their only motherboard to support DDR3 during that time. How many pins is DDR4 though? Still 240? About 90$-100$ for 8GB of DDR4 when ready for consumer market, expected to drop to current DDR3 prices after 2 years. Is that a normal cycle? Technology transitions so much faster these days.
DDR4 debuting at under $100 is very unlikely.
|
+ Show Spoiler +On November 18 2013 04:55 Kojak21 wrote:
What is your current build?
CPU: intel core i5 2500k
GPU: Nvidia GeForce GTX 560 TI
8 Gigs of ram
1 TB Hdd
What is your monitor's native resolution?
1920 x 1200
Why do you want to upgrade? What do you want to achieve with the upgrade?
I figured i would treat my self this Christmas with something new for the computer. I'm looking for something that will basically asking if there are any weak points to my system that could use upgrading.
Ive been looking into those SSD things and that seems like a good choice to me, but I'm not that computer knowledgeable. any input from someone who knows more would be awesome.
What is your budget?
no more than 500
What country will you be buying your parts in?
Canada
If you have any brand or retailer preferences, please specify.
Doesn't really matter
-
If i missed anything important out let me know!
thanks
Just one bump for me cause i never really got an answer
|
Well it depends on what you want to achieve? If you want better FPS in the latest AAA titles than upgrade your video card. If you want a more responsive system than grab an SSD.
|
On November 20 2013 08:24 skyR wrote:Show nested quote +On November 20 2013 05:09 Disregard wrote:On November 20 2013 00:00 iTzSnypah wrote:
DDR4 is going to be crazy expensive and everyone is going to stay with DDR3 until prices become reasonable (just like when DDR3 came out, it wasn't fully adopted for over 2 years). My motherboard was MSI's flagship when DDR3 came out and it was their only motherboard to support DDR3 during that time. How many pins is DDR4 though? Still 240? About 90$-100$ for 8GB of DDR4 when ready for consumer market, expected to drop to current DDR3 prices after 2 years. Is that a normal cycle? Technology transitions so much faster these days. DDR4 debuting at under $100 is very unlikely.
My guess is at least $250 for 16GB (the minimum). Quad-channel = need 4 sticks and I highly doubt they are going to make 2GB sticks.
|
On November 20 2013 09:00 Kojak21 wrote:+ Show Spoiler +On November 18 2013 04:55 Kojak21 wrote:
What is your current build?
CPU: intel core i5 2500k
GPU: Nvidia GeForce GTX 560 TI
8 Gigs of ram
1 TB Hdd
What is your monitor's native resolution?
1920 x 1200
Why do you want to upgrade? What do you want to achieve with the upgrade?
I figured i would treat my self this Christmas with something new for the computer. I'm looking for something that will basically asking if there are any weak points to my system that could use upgrading.
Ive been looking into those SSD things and that seems like a good choice to me, but I'm not that computer knowledgeable. any input from someone who knows more would be awesome.
What is your budget?
no more than 500
What country will you be buying your parts in?
Canada
If you have any brand or retailer preferences, please specify.
Doesn't really matter
-
If i missed anything important out let me know!
thanks Just one bump for me cause i never really got an answer
Check out Sandisk Extreme II 240GB on Saturday at memory express. I am hoping for $150.
|
See,THIS is what I fear the most, what I can't understand. How the hell am i suppose to know what to buy?!
|
Why would you buy from Best Buy?
Well one clearly stands out by having a different heatsink (single fan, blower style), you don't want that one. The other two differs in clock speed which isn't a big deal, you can adjust clock speed yourself with software in Windows. If you want a guarantee overclock for whatever reason than you can pay a premium to have that guarantee.
|
I was looking around, starting to compare prices to get an idea of what is a good deal or not. But then I see 2 identical items, with a significant price difference with no idea, whatsoever, what is different and if it matters.
I been studying the first post to know what I was getting into but there are all those tiny little information that you just need to ... already know?
|
United Kingdom20343 Posts
The other two differs in clock speed which isn't a big deal, you can adjust clock speed yourself with software in Windows. If you want a guarantee overclock for whatever reason than you can pay a premium to have that guarantee.
It's a big deal on 770's where a few models of cards have volt unlock but most don't, and negligent of them not to list the clockspeeds or other info so that you know which is which. Classy vs standard may even have different pcb etc
|
Not really. If you're a user that cares about volt unlock than you'd do research on which is volt unlocked. All of those EVGA's are nothing special, it's only Classified and FTW that uses custom afaik.
|
$400 a 2 Gb 770? Isn't that a bad deal?
http://www.newegg.ca/Product/ProductList.aspx?Submit=ENE&DEPA=0&Order=BESTMATCH&Description=gtx 770&N=-1&isNodeId=1
I'm sure that there's a 770 with voltage unlocked in that list at ~$320 (ah shipping though).
Anyway I'm on a quest to learn more about CPUs in general. big post in spoiler, many questions asked:
+ Show Spoiler +Allow me to examine intel's specifications on its i5 4670: http://ark.intel.com/products/75047/So there are a couple numbers that speak for themselves. Clock speed @ 3.4 GHz: the rate at which the processor works. the higher the better, obviously Lithography @ 22 nm: the way I would put it, the lower this number, the more transistors can be put into the CPU, meaning it does more work per clock cycle. however, it also depends on architecture, or how everything in the CPU itself is layed out. I don't know much about this specifically. Max TPD @ 84W: we went over this with Cyro recently. Basically the amount of heat in watts that the CPU will be generating, at full load. or is it maximum power draw and a fraction of that 84w is made into heat? I imagine this number gets higher if we get a 4670k and overclock it. Cache @ 6 Mb: This is kind of like RAM except for a processor? I thought there were three levels of cache though. first level is very small but very fast (for executing "small" programs?) second is a bit bigger and slower and third is biggest and slowest. so basically if we ask the processor to execute a small program, it'll do that "in" the the smallest cache and it'll do that very quickly. the larger the task we ask of it, the more cached is used. however 6 Mb is really small so rather than seeing it as RAM, isn't it more like a kind of like.. the amount of data which can be processed by the CPU in a given time? so the higher the cache, the more data can be processed at a time, the "faster" or more "powerful" the processor? meh I have no idea what cache is. besides all these numbers, what else characterizes a CPU? what should I be looking for when I'm trying to figure out if AMD's new stuff is great or not?
|
United Kingdom20343 Posts
TDP stands for Thermal Design Power and can mean lots of things, not sure on specifics. In both current Intel cpu's though and Kepler GPU's, there is a sensor that views power numbers and controls clocks, a 4670k will go up to max turbo bins as long as it reads below 84w (or whatever you set) but if it hits it, it will throttle clocks, same with a kepler GPU, they'll boost up to maximum clock speed value unless they hit a temperature limit (which won't really happen because the gpu's run cool unless you're oc'd with a 780+) or until they're power throttled, with kepler cards this is even expressed in % of TDP
besides all these numbers, what else characterizes a CPU? what should I be looking for when I'm trying to figure out if AMD's new stuff is great or not?
Too much stuff, looking at alone. Much simpler to make comparisons and look at what is changing and why. Without a similar piece of hardware to compare it to, performance estimates are wild at best
|
I'm looking for a GPU upgrade, currently on my mind is nothing big but these three cards, GTX 670/760 and HD 7950, here in UK all of these go around for £190-200 on scan.co.uk, so the price is not a factor here, checking reviews, benchmarks, they all seem pretty even, convince me to one, alternative is HD 7970 for for £230, might be worth the investment?
|
A 7950 PCS+ around 200€ or a 280x/GTX770 around 250€ are pretty good value at the moment
|
On November 20 2013 14:52 Incognoto wrote:
Lithography @ 22 nm: the way I would put it, the lower this number, the more transistors can be put into the CPU, meaning it does more work per clock cycle. however, it also depends on architecture, or how everything in the CPU itself is layed out. I don't know much about this specifically.
It has to do with the the spacing between components on the board, essentially you could say that at the sub-component level, a typical subcomponent or "building block" would be 22nm across, ie. a transistor. These components are all sort of molded onto a silicon wafer with various photolithography techniques, to achieve the desired outputs to various inputs (typically electrical). Process nodes getting smaller means that the bridging connections use less power and may have more throughput in certain critical areas, however the leakage of each small "building block" increases with the small size, and that can quite significantly hinder shrinkage of a node. See Intel's tri-gate technology to try to reduce gate leakage in transistors.
Process node size has nothing to do with IPC (instructions per clock), this has to do with the how the different logic units of the cpu are designed, largely the main pipeline among other things. If you really want to learn about cpu architecture it will take a fair while to have even a basic understanding. I've done a basic course on building a cpu from boolean logic, and i can assure you it was not fun.
On November 20 2013 14:52 Incognoto wrote:
Max TPD @ 84W: we went over this with Cyro recently. Basically the amount of heat in watts that the CPU will be generating, at full load. or is it maximum power draw and a fraction of that 84w is made into heat? I imagine this number gets higher if we get a 4670k and overclock it.
Do you know about the law of conservation of energy? There are a few fundamental forms of energy, and ignoring the energy to mass conversion, it can't ever be "lost". Light, heat, sound, kinetic, nuclear, elastic, chemical etc. As you can imagine, your processor is basically silent, doesn't really move much, and certainly doesn't light up much, hence basically <99% of the energy "consumed" by the processor is converted to heat. Heat increases with increased power draw, and P = V^2/R, so yes, power would increase (exponentially) as voltage increases (for most loads), and voltage increases are required for overclocking.
On November 20 2013 14:52 Incognoto wrote:
Cache @ 6 Mb: This is kind of like RAM except for a processor? I thought there were three levels of cache though. first level is very small but very fast (for executing "small" programs?) second is a bit bigger and slower and third is biggest and slowest. so basically if we ask the processor to execute a small program, it'll do that "in" the the smallest cache and it'll do that very quickly. the larger the task we ask of it, the more cached is used. however 6 Mb is really small so rather than seeing it as RAM, isn't it more like a kind of like.. the amount of data which can be processed by the CPU in a given time? so the higher the cache, the more data can be processed at a time, the "faster" or more "powerful" the processor?
meh I have no idea what cache is.
You are right, most cpus these days have three levels of cache. L1 holds the immediate currently used/just used variables (that aren't in the cpu registers, which are the things that the memory is loaded into), what is stored depends on OS scheduling. L2 is fair bit slower, and L3 is much slower and larger, and usually contains frequently used data, maybe some variables from half a second ago, larger constructs, etc.
The reason there are different levels of cache is because in order to get really fast access to the cpu, it needs to be close, which means small. It's also extremely expensive to try to make a lot of very very small fast access memory. If we could have a 6MB L1 cache we would, and it would be amazing  , not to mention make performance programming much easier.
Typically, access speeds might be something like L1 = 1x, L2 = 3x, L3 = 10x, Memory = 50x, harddrive = 1000+x or something, (exact numbers are completely out of my ass but they give an idea of how important cache over memory is), you don't want to be putting intermediate variables straight from registers into memory and pulling it out again, you'd be chugging so slowly it's not funny.
On November 20 2013 14:52 Incognoto wrote:
besides all these numbers, what else characterizes a CPU? what should I be looking for when I'm trying to figure out if AMD's new stuff is great or not? [/spoiler]
Basically, benchmarks.
Sorry for the late reply, I started writing this and forgot about it for a few hours  .
|
|
|
|
|
|
|
|
EPT
