- Status
I believe I get away cheaper than if I were buying a console, these days. Primarily because of Sales and quick decline of prices.
With very few exceptions, I don't spend more than €15 per game (most often less than €10). Having a big backlog and no need to buy any specific game at any specific point in time helps. Consoles, as far as I'm aware - although I don't know what it looks like with the upgraded online services of the new generation - hardly ever get to those prices. It's a rarity retail console games get below €20 - and that's years after release. Even if I buy brand new games, I likely get it for €45 - €50, rather than €60 - €70 for a console game.
So I save at the very least €5 per game, more likely about €20 to €30 if I were to buy the console game at the same time as I tend to get the PC game. Add that about everyone needs a PC or tablet or whatever for office stuff and my gaming rig doubles as that. If someone were to have a computation intense hobby - like 3D art, making videos, etc - that someone even would have to go for a decently powerful CPU, quite a bit of RAM and often a somewhat fast GPU for DirectCompute/CUDA.
Since you are the expert please correct me if I am wrong on this but...
Obsolescence for gaming? Cause that's what we're talking about here after all. As far as I know an I7 3770k Ivy Bridge will run games with very similar performance to the latest Intel CPU, hell even a sandy bridge one could do it so what's the point in upgrading the CPU when it would involve such annoyances as changing your motherboard to be able to make full use of your new hardware?
I've had my Ivy Bridge for over 2 years, getting close to 3, and I see no real reason to change it for at least 2-3 more years at the rate of how it looks for gaming and when I do replace it I want to sell and so overclocking for the mild performance increases is just not worth the heat, the noise, the risk you're taking and the cost of cooling. ( Though frankly I can crank my fans up to max and crank it above 4.3 ghz and maintain good temperatures since I've tested that ).
It's worth it far more to upgrade my GPU, hell even go SLI if I want real meaningful performance increases. With my current GTX 780 I can achieve 1080p/60 FPS+ in most games on their highest settings and even those that I can't I just tone down the AA and maybe a few other things and I'll get there. Once I get a GTX 980 I'll play virtually every game on the highest possible setting at 60 FPS 1080p unless that game is optimized like shit and anything above that in resolution/FPS is just a luxury few can afford/care for. Though if I wanted that I'd go yolo and get a Titan X.
Last edited:
Not likely...Prepare yourself for a slideshow
I think Guy's obsolescence is, in this case, referred to the amount of time a CPU is useful for gaming. Not just general usage. And even though this time is getting longer than before (compare Pentium 2 to 4, and sandy bridge to haswell) it is still measured in one digit years. The point being, you'll replace your CPU before you degrade it from overclocking.
But yes, each generation games are more dependant on GPU than anything else, and even a 4 year old CPU can satisfactorily drive a modern GPU.
But yes, each generation games are more dependant on GPU than anything else, and even a 4 year old CPU can satisfactorily drive a modern GPU.
Last edited:
Agreeed, CPUs can continue to be useful for more than just one or two tick-tock cycles. But that does not lead to the conclusion that overclocking a CPU is likely to reduce its service life to a time less than its obsolescence.
The extra power consumption, heat, cooling requirement, noise, void warranties, and general hassle are valid arguments against overclocking if these are nuisances to you. But I want to see evidence that properly executed overvoltage and overclocking reduces the service life of any CPU newer than a bleeding Northwood before coming down against the practice.
REDpoints are the same as "likes" on other forums. The forum keeps a count of REDpoints you have received.
I like to use "thanks" to mean "that was a useful solution to a problem, or a fine contribution to a discussion", and REDpoints to mean "that really touched me. I learned something about you or about myself from that post".
Certainly some posts qualify for both.
and nice to meet you
---------- Updated at 10:44 AM ----------
and i am iranian
---------- Updated at 10:44 AM ----------
and i am iranian
Fair enough.
On another subject I'm going to change my regular Windows HDD for an SSD. Currently thinking about this one: http://www.kingston.com/en/ssd/v#sv300s3. I do wonder if it's worth it though, not price but performance wise. Storage wise I don't need more then 500 GB because I do already have two 120 GB SSDs.
On another subject I'm going to change my regular Windows HDD for an SSD. Currently thinking about this one: http://www.kingston.com/en/ssd/v#sv300s3. I do wonder if it's worth it though, not price but performance wise. Storage wise I don't need more then 500 GB because I do already have two 120 GB SSDs.
On the subject of overclocking, my thoughts are DO IT! It's free performance, and if you do it right, the lifespan of the CPU will not be significantly diminished unless you plan to keep it for 10 years or so, by which point it will be obsolete.
The thing I've learned about overclocking, is that every CPU has a sweet spot. The sweet spot being the smallest voltage increase above default, to get the highest and most stable clock frequency. For my CPU which is a 4930K, it's 4.3ghz. To hit 4.3ghz and be rock solid stable, my CPU only requires 1.21v (using offset voltage), which is nothing. I can also hit 4.5ghz at 1.285v, but to me the rise in voltage wasn't worth the minimal performance gain of 200mhz..
Before I had my current CPU, I had a 3930K which I had running at 4.5ghz at 1.365v. It was stable for about a year, then I started getting BSODs. I had to increase the voltage to remain stable at 4.5ghz, and eventually I needed 1.4v to do so, which was when it finally dawned on me that I had ruined my CPU by pumping too much voltage into it and accelerating electromigration.
So in the end, I sold that 3930K to someone who I know doesn't overclock for $350 and bought the 4930K and vowed never to make that mistake again But overclocking is definitely worth it though, as it can dramatically reduce a CPU bottleneck..
The thing I've learned about overclocking, is that every CPU has a sweet spot. The sweet spot being the smallest voltage increase above default, to get the highest and most stable clock frequency. For my CPU which is a 4930K, it's 4.3ghz. To hit 4.3ghz and be rock solid stable, my CPU only requires 1.21v (using offset voltage), which is nothing. I can also hit 4.5ghz at 1.285v, but to me the rise in voltage wasn't worth the minimal performance gain of 200mhz..
Before I had my current CPU, I had a 3930K which I had running at 4.5ghz at 1.365v. It was stable for about a year, then I started getting BSODs. I had to increase the voltage to remain stable at 4.5ghz, and eventually I needed 1.4v to do so, which was when it finally dawned on me that I had ruined my CPU by pumping too much voltage into it and accelerating electromigration.
So in the end, I sold that 3930K to someone who I know doesn't overclock for $350 and bought the 4930K and vowed never to make that mistake again
But overclocking is definitely worth it though, as it can dramatically reduce a CPU bottleneck..Performance is measured in operations or products per unit of time, not clock frequency which says next to nothing about it.
We talked about bottlenecks before. Different CPUs behave differently and if one gives you better performance with a given GPU, it doesn't necessarily mean the other is "bottlenecking" it. A bottleneck is a visible limitation of maximum performance.
Exactly how much faster do games get after a CPU overclock? On average, because some games are completely unaffected. Dramatic is a heavy word and while I understand this is a gaming (not a technical) community, I think we should be more educated posters.
We talked about bottlenecks before. Different CPUs behave differently and if one gives you better performance with a given GPU, it doesn't necessarily mean the other is "bottlenecking" it. A bottleneck is a visible limitation of maximum performance.
Exactly how much faster do games get after a CPU overclock? On average, because some games are completely unaffected. Dramatic is a heavy word and while I understand this is a gaming (not a technical) community, I think we should be more educated posters.
The CPU performance metric (IPC) is measured by instructions per cycle multiplied by clock speed, so clock speed is obviously an important factor in CPU performance.
Source
A CPU bottleneck exists when the CPU isn't fast enough to give the GPU(s) draw calls and the GPU ends up waiting.. We want the GPU to be functioning at maximum performance, and it cannot do so if the CPU isn't fast enough to keep up with it. That said, lots of factors can influence bottlenecks and not just hardware. Settings has a big influence on determining whether a game is CPU or GPU bottlenecked as well..
This question is too nebulous to answer properly without some context. For someone like me with SLI, CPU overclocking can increase performance substantially in games that use a single thread for draw call commands. In multithreaded games, the performance increase would be smaller as multiple threads are uploading information to the primary thread to send to the GPU, but there would still be a performance gain.
Here is a review which explores the impact of CPU overclocking on performance in games with single GPU and SLI.
Instructions per cycle is not an actual metric of performance. But you should read the rest of that wikipedia page:
"instructions per clock is not a particularly useful indication of [the] performance".
It is a factor in determining measurable performance, but not a metric. If you want to measure and compare computer performance, use FLOPS or in the case of games (a very applied scenario), FPS under controlled scenarios.
what i meant to say before is "bottleneck" is a buzzword used way too often. Sure you may improve performance by overclocking it, but unless you prove a given CPU is unable to increase game performance with varying, more powerful video cards, you can't really say it's bottlenecking the system, i.e. fixed maximum throughput.
Your link is interesting and somewhat useful, but it's measuring SLI scaling (not directly OC). We can use their minimal testing scenarios to draw some conclusions though:
Performance increase with a 780 SLI after a 40.63% increase in CPU clock frequency.
1) Some games that are probably not (properly) multithreaded:
Skyrim:
+22.3% FPS / Low detail and res
+21.77% FPS / Max
Battlefield 3:
+5.93% FPS / Low detail and res
+1.26% FPS / Max
2) Now some games that are probably multithreaded:
Metro Last Light:
+18.57% FPS / Low detail and res
+5.26% FPS / High
Bioshock Infinite:
+25.81% FPS / Low detail and res
+28% FPS / Max
So after a 40% bump, the best we see is a 28% increase in actual game performance. Judging by that review you suggested, it's not such a great thing really. In parallel computing we measure efficiency, a unit of resource utilization, as the ratio of speedup and processors (or resources). If the resources increase faster than the speedup, it is evident the efficiency decreases and tends to zero as the amount of resources increase. This is actually the case here. so while a rare maximum of 28% might be worth it gamer-wise, it's somewhat inefficient actually. At least with their test setup.
Last edited:
You should have pasted the entire sentence, and not just some of it. It said:
"For users and purchasers of a computer system, instructions per clock is not a particularly useful indication of the performance of their system."
The context of that sentence is that for end consumers, IPC is not a useful indication of the performance of their system, and I would agree because most consumers are clueless about such things. Using pretty graphs and benchmarks to illustrate performance is much more useful, as it's easier to understand and more relevant to a particular workload. .
This doesn't mean however that IPC is not useful. IPC is still used by engineers and professionals to this day to gauge a CPU's performance.. When Soft Machines unveiled their VISC CPU architecture, they used IPC to compare the performance of their new architecture against other contemporary designs..
Source
I never used the word "bottleneck" ambiguously if you recall. I specifically said CPU bottleneck. Lots of things can cause bottlenecks in a system I agree, but when I say CPU bottleneck, thats exactly what I mean..
I think it's more useful to disregard the SLI benchmarks, and focus on single card because SLI presents more difficulties due to SLI scaling issues and drivers and what not. BF4 performance for example has increased significantly since that article took place due to patches, SLI profile and driver updates.
Anyway, you missed the greatest example of a performance increase due to CPU overclocking. Sleeping Dogs got a massive 70% increase in performance from overclocking the CPU, which seems almost unbelievable... Bioshock Infinite got a 32% increase at max detail in single card mode.
Last edited:
I don't know about CPU lifeor GPU life or what but I can tell you I live in Australia and it gets to 45 degrees here and my liquid cooling system is fucking ace --- computer never gets higher than 50 degrees, on a normal day usually 35~. if you don't touch the unit, it isn't going to leak. it's designed not to. you should never have to open the unit for any reason whatsoever and if you do, whoever sold it to you needs to be hung out to dry.
All I can say is that it is extremely good if you live in a country made of fire. I cannot speak about any other effects than that it's basically essential if you want to game during the daytime in summer in a desert.
I'm ready to end the IPC argument but it should be said that document where they refer to instructions per cycle is not technical, it's PR. And while you can measure anything and see how it affects the behavior of things, IPC is not useful in your specific case of overclocking because it's a design feature. But yes it's a better indicator of expected performance than, say, clock frequency.
Again, a real measure of performance is for instance FLoating point Operations Per Second, specially because it's architecture independent. And games aren't benchmarks, often they're poorly designed and scale awkwardly. In other words, it's hard to justify OC as having a substantial performance effect in real life scenarios (gaming). That's not saying it doesn't help, it does. Just that using proper metrics, is not very efficient and therefore hardly a "massive improvement".
I'm not against OC. I think it's a fun project and you get to learn some elementary CPU concepts. But it's also not such a fantastic, amazing, life changing experience.
Again, a real measure of performance is for instance FLoating point Operations Per Second, specially because it's architecture independent. And games aren't benchmarks, often they're poorly designed and scale awkwardly. In other words, it's hard to justify OC as having a substantial performance effect in real life scenarios (gaming). That's not saying it doesn't help, it does. Just that using proper metrics, is not very efficient and therefore hardly a "massive improvement".
I'm not against OC. I think it's a fun project and you get to learn some elementary CPU concepts. But it's also not such a fantastic, amazing, life changing experience.
Last edited:
- Status