Myths and other old wives tales regarding PC hardware

PsychoFish

PsychoFish

New member
Ok, so after wanting to smash someone's head though a brick wall I'm posting this to ensure more people know the real differences and we can do some Mythbusting®

1st on my list Intel Xeon processors are faster than Intel Core i5/i7 processors

Quick answer : BUSTED, they use the same underlying architecture

Long answer :

Although they are fundamentally the same
  • TDP (Thermal Design Power) on the Xeon is lower than that of a comparable i5/i7
  • Xeon processors do not ship with embedded Intel GPUs
  • Xeon processors support multi-socket configurations
  • All Xeon processors are HT (HyperThreading) enabled
  • Xeons are designed to run at high load for extended periods of time
  • The Xeon processors CANNOT BE OVERCLOCKED
  • Xeons have much, much more L3 cache than a comparative i5/i7
  • Only Xeon processors support ECC RAM

So yes, if you want a CPU than can run at 80+% load 24/7 at a relatively decent temperature the Xeon is for you. But in terms of pure bang for your buck (performance/$) the i5/i7 is the way to go.

Myth 2: ECC RAM (Error-correcting code memory) is better than non-ECC RAM

Fact: ECC RAM at the same clock speed as non-ECC RAM is actually slower
Fact: ECC RAM is more expensive than non-ECC RAM
Fact: ECC memory is used in most computers where data corruption cannot be tolerated under any circumstances, such as for scientific or financial computing.

Is it better? Yes?
Do you want in in a gaming PC? No
Do you need it for Excel? No
Do you need it at all in your workstation? Not unless the work you're doing requires massive amounts of in memory number crunching which requires an extremely high level of accuracy.

Myth 3: AMD CPUs are slow

NOPE NOPE NOPE, AMD CPUs are not slow, not at all. In pure terms of performance/$ the AMD wipes the floor with an i7. The real issue as to why they can be perceived as slow is a fundamental architecture/design issue which can cause the CPU to under perform under certain conditions where applications heavily rely on multi-core processing.
 
Myth: PC Cd rom drives topped out at 56x speed

Kenwood designed a 72x drive using multiple optical beams, increasing throughput up to 72× (TrueX technology) with a 10× spin speed, but along with other technologies like 90~99 minute recordable media and "double density" recorders, their utility was nullified by the introduction of consumer DVD-ROM drives capable of consistent 36× CD-ROM speeds (4× DVD) or higher.

72x_box.jpg
 
Myth : Thermal paste makes a huge difference to the cooling of your CPU/GPU temperature

Reality, YES & NO

A good quality thermal paste will ensure that better heat transfer occurs between the chip and the heatsink. The better the thermal conductivity of the paste the more efficient it will be. Actual tests have shown that the temperature variance between pastes using similar tests and the same CPU/GPU and cooler results in a spread of 5 degrees C...yes that makes the difference between 95 and 100 degrees, but in reality you're better off getting a better cooling device/heatsink than you are changing your thermal paste

Hell, tooth paste (for reference http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616-19.html) is only 10 degrees more than the crappiest thermal paste.
 
[MENTION=6600]PsychoFish[/MENTION]: With respect to what you said about AMD processors, do you mean that some benchmarks should be taken with a pinch of salt? Are the real-world differences perhaps not as glaring as review results would suggest?

Thank you
 
phoenix said:
[MENTION=6600]PsychoFish[/MENTION]: With respect to what you said about AMD processors, do you mean that some benchmarks should be taken with a pinch of salt? Are the real-world differences perhaps not as glaring as review results would suggest?

Thank you
Click to expand...

Correct, however don't expect your AMD FX-9590 to be an Intel i7 4770K and you'll be fine. While the Intel i7 is still THE processor to have you can get away with a FX-9590/FS-8350 without being that much slower than say an i5 4670K.

So to say the AMDs are slow is a relative statement. I like using cars for comparisons. It's like buying a BMW M3 and saying it's a shitty bulldozer. If you wanted a bulldozer you should have bought one in the first place. The AMD CPU is a BUDGET CPU, comparing it to a premium CPU is just lunacy.

However if money is a huge factor and you still want to play the latest and greatest, then the AMD is acceptable. As long as you realize you bought the Polo GTi of CPUs you'll be ok.
 
Cranky said:
@Pshychofish i need your help to understand something . Why is Overclocking I7 4970 and 6700K not helping your performance in game?

http://www.anandtech.com/show/9533/intel-i7-6700k-overclocking-4-8-ghz/8
http://www.youtube.com/watch?v=6bjrKFiIWlU
Click to expand...

Simple, we're not in the days of single core dominance anymore. Overclocking is only increasing your CPU's clock speed.
While clock rates are a valid way of comparing the performance of different speeds of the same model and type of processor, other factors such as pipeline depth and instruction sets can greatly affect the performance when considering different processors. For example, one processor may take two clock cycles to add two numbers and another clock cycle to multiply by a third number, whereas another processor may do the same calculation in two clock cycles. Comparisons between different types of processors are difficult because performance varies depending on the type of task. A benchmark is a more thorough way of measuring and comparing computer performance.

Essentially the CPU is already running as efficiently as possible and making it execute a minor amount of additional calculations per second will make less of an impact than increasing the number or cores.

Basically think of this : Cores > HT > Clock Speed

Also keep in mind that the newer games are taking better advantage of multi-core CPUs.
 
Myth: if a PSU says its a 500w PSU, it automatically means it can handle gaming rigs.

FACT: PSU's come in all shapes and sizes, they are not created equal. A cheap 650W PSU might actually be worse-off (in actual power, lets not even get started on efficiency, clean output etc, ill leave that to [MENTION=6600]PsychoFish[/MENTION], who explains these things better) than a good quality 450W PSU.
Too many people look at a PSU and go "well it doesnt effect my FPS performance in bechmarks, so no care!" and then go buy the cheapest PSU they can find that has an adequate looking number...
This makes me facepalm...

Myth: A 1000W+ PSU is overkill and a waste of electricity.
FACT: A high wattage PSU will actually SAVE you money. Heres why: PSUs work best at "average" (50%) load.

That means if you expect your "normal" usage to be 500w, you should be buying a 1000w.

PSUs always have their maximum efficiency in the 40 - 60% range so the 500w on a 1000w PSU gives you the biggest bang for your bronze/silver/gold/platinum buck.

A 1000w platinum running at 50% would be about 90% efficient which means that 500w internal would be pulling about 550w from the wall and only 50w would be waste heat inside the PSU.

On the other hand, if you had a 500w platinum PSU pulling 500w (100% load) then your efficiency would likely be about 80% and you would be using 600w from the wall with 100w being heat inside the PSU (100% more).

Link here with better wording and pretty pictures.


EDIT: added link
 
Last edited:
mottamort said:
Myth: if a PSU says its a 500w PSU, it automatically means it can handle gaming rigs.

FACT: PSU's come in all shapes and sizes, they are not created equal. A cheap 650W PSU might actually be worse-off (in actual power, lets not even get started on efficiency, clean output etc, ill leave that to [MENTION=6600]PsychoFish[/MENTION], who explains these things better) than a good quality 450W PSU.
Too many people look at a PSU and go "well it doesnt effect my FPS performance in bechmarks, so no care!" and then go buy the cheapest PSU they can find that has an adequate looking number...
This makes me facepalm...

Myth: A 1000W+ PSU is overkill and a waste of electricity.
FACT: A high wattage PSU will actually SAVE you money. Heres why: PSUs work best at "average" (50%) load.

That means if you expect your "normal" usage to be 500w, you should be buying a 1000w.

PSUs always have their maximum efficiency in the 40 - 60% range so the 500w on a 1000w PSU gives you the biggest bang for your bronze/silver/gold/platinum buck.

A 1000w platinum running at 50% would be about 90% efficient which means that 500w internal would be pulling about 550w from the wall and only 50w would be waste heat inside the PSU.

On the other hand, if you had a 500w platinum PSU pulling 500w (100% load) then your efficiency would likely be about 80% and you would be using 600w from the wall with 100w being heat inside the PSU (100% more).

Link here with better wording and pretty pictures.


EDIT: added link
Click to expand...

While your facts are correct, I've got a hunch that your conclusion might be wrong. Give me a couple of minutes to excel up some numbers.
 
Last edited:
Okay, everyone, spreadsheet time! \:D/

Some disclaimers first:

1) I got all my data from Tom's mother flippen' hardware, Pricecheck.co.za and a Tshwane based prepaid electricity reseller. Whose name I can't remember now.

2) I didn't cherrypick the PSU's; I took two from the most efficient end of the spectrum, two from the least efficient end, and one right in the damn middle. (At the mean efficiency, for the stats nerds, not the median) This means that my expensive ones might be overly luxurious and makes you cocktails while you game, and the cheap ones excessively peasant-like, giving you black lung. I did not research the PSU's at all, and if I had I certainly would've had less trouble with point 3, which is..

3) With hindsight, I should first have checked if I can find a price for each PSU locally. Pulling the data took a couple of minutes, since I id it manually, so I really didn't feel like doing it over when I found that 3 of the 5 PSU's I picked did not have a local price on Pricecheck. Poop. I also would've been more careful to picked ones so that we can utilize the 40% - 60% band more, which slightly invalidates my efforts. But I already committed in my last post. Like a fool.

I'll gladly upload my spreadsheet if someone else want to play with it, or add more PSU's or colour it in, or wave it in someone's face or print it out and feed it to your dog because you don't like me/PSU's/Excel/yourself/your dog/Heyneke Meyer, or make paper aeroplanes.

PcvPKJK.png


As I mentioned above, I picked 5 PSU's, and 3 load cases: 550, 300, and 150 Watts, which I believe are decent numbers for gaming, working, and idling, respectively. I used linear interpolation to estimate the efficiency for each load case, which which I calculated the total power draw. Subtract the useful power, and you get the wasted power (Which [MENTION=12607]mottamort[/MENTION] correctly tells us becomes heat, but so does all of the power going into your pc. It's a state machine, no work [in terms of physics, not MS Office] is being done, so all that power becomes heat, with the exception of the minuscule amounts required to spin the fans and old school hard drives). Wasted power is expressed in Watts, while we pay per kWh used. The cost of wasted power is the wasted power in Watts, multiplied by the cost in Rand per kilowatt-hour multiplied by the time the machine ran in hours, divided by 1000 to correct for the kilowatts not being Watts.

It's immediately clear that running the 550 W PSU at 100% load costs (in wasted power) nearly double running the 1500 W PSU, at 14 c per hour vs. 8 c per hour. (These are all rounded to the nearest cent, or two significant decimal places, but the ~32bit precision of Excel carries over through the calculation, even when it looks rounded). However, there is a massive difference in cost.. A staggering R 5821 for the 1500 W PSU compared to R 643 for the 550 W. That's pretty much 512 Gb of SSD harddriveness!

The amount that you'd have to shell out more for your 1500 W PSU is R 5178. The efficiency gains and savings for using this PSU converts to 7.64 years of continuous running of your PC at load in order to break even.

Conclusion: It's not worth the extra cash. I might well be proven wrong by having used a ridiculously overpriced PSU made from diamonds and Nkandla grass, and I'll gladly revise the spreadsheet if anyone points me to a specific PSU; I'm not going to go look for one though, since I like my conclusion. Because I wrote it.
 
Avatar said:
~snip~
Click to expand...

You said a couple of minutes! Unacceptable :D

I agree though. It is like buying a diesel car because it is more fuel efficient, so you will be saving on fuel costs. However the diesel version of the car is R20 000 more expensive than the petrol car.

You'd have to drive the diesel car for quite a long time (put a lot of kilometers on it) to make up that R20 000 and to actually start saving money (not even taking interest into account). A friend and myself once did the math, but I can't recall what we conclusion we came to, but it was not practical at all.
 
Blazzok said:
You said a couple of minutes! Unacceptable :D

I agree though. It is like buying a diesel car because it is more fuel efficient, so you will be saving on fuel costs. However the diesel version of the car is R20 000 more expensive than the petrol car.

You'd have to drive the diesel car for quite a long time (put a lot of kilometers on it) to make up that R20 000 and to actually start saving money (not even taking interest into account). A friend and myself once did the math, but I can't recall what we conclusion we came to, but it was not practical at all.
Click to expand...

My dad and I did a similar calc on my first car. We decided that at 36km travelling to the university and back it wouldn't make sense to get the diesel. However now seeing as the rand is shot, it's cheaper to cheaper to buy booze than petrol and with everything else getting so damn expensive I wish we got the diesel.

Obviously it's harder to predict inflation and the influence of the Zuma squad but looking back, I think I'll take the day 1 knockout as apposed to the continual throb of Zuma tapping on wallet. I think a similar thing could be applied here. Granted you may only need a 600w psu but hurt yourself a little and get a 800w psu. It's less taxing on your rig and it gives you a little extra breathing room in the uncertain future.
 
Ike_009 said:
My dad and I did a similar calc on my first car. We decided that at 36km travelling to the university and back it wouldn't make sense to get the diesel. However now seeing as the rand is shot, it's cheaper to cheaper to buy booze than petrol and with everything else getting so damn expensive I wish we got the diesel.

Obviously it's harder to predict inflation and the influence of the Zuma squad but looking back, I think I'll take the day 1 knockout as apposed to the continual throb of Zuma tapping on wallet. I think a similar thing could be applied here. Granted you may only need a 600w psu but hurt yourself a little and get a 800w psu. It's less taxing on your rig and it gives you a little extra breathing room in the uncertain future.
Click to expand...

Hold your horses, there's more to PSUs that just wattage. I'll post something more in depth on Monday.
 
You must log in or register to reply here.
Back
Top