Retailer reveals most reliable graphics cards and motherboards, warranty process times

[alexnode]

I had quite a few Palit cards over the years, I stopped using Asus motherboards and switched to Supermicro. ... EVGA, silverstone for PSUs and crucial/micron for memory. In general if you want something good make sure it hasn't got any RGB saying gaming on it but at the same time it's not the cheapest brand available.

 

[AdamNovagen]

You want to know what irritates me about the Steam Survey? The fact that I've never managed to get it to work on ANY of my PCs and I've tried for YEARS on three separate platforms and four different motherboards. It makes me wonder just how many other PCs with Radeons don't get counted. Don't be so sure of yourself when you refer to the Steam Survey as accurate because Steam has been aware of the issue but has done nothing to address it. If it doesn't work for all PCs, it's not accurate, period.

You are actually wrong about this. Random sampling is a well-proven methodology in statistics, and it's actually explained fairly well in this reddit post about the July 2021 SHS:

Myth - The Steam Hardware Survey is inaccurate because I didn't get polled. or They have to poll everyone for it to be accurate.



The SHS relies on Random Sampling, which is a well understood practice in statistics.

As a basic example; Imagine that you had a population of squirrels, let's just say 5000, and you wanted to know how many squirrels had grey fur, versus brown fur. You don't have to capture all 5000 squirrels to figure out how many squirrels have each color of fur. You can instead randomly sample a few hundred squirrels, and figure out a ratio, or percentage, of that smaller random sample pool that have each color of fur. And then apply that to the larger population to get a fairly accurate estimation of the color distribution of the squirrels fur.

You have to make sure your sample size is large enough to not introduce sampling errors, but if you do have a large enough sample size, you can get a very representative breakdown of data.

Random sampling works quite well, as long as you get a good sample size. According to this article on the subject, a sample size should be minimum 100, maximum 10% of the total population. They cap this at 1000 individuals, and the SHS definitely surveys way more than that, so you can rest assured it can be expected to be accurate to within 3% up or down. And you know what, AMD cards have expanded a bit in the latest surveys anyway, so I'm happy enough lol

 

[captaincranky]

In general if you want something good make sure it hasn't got any RGB saying gaming on it but at the same time it's not the cheapest brand available.

Well, the fact of the matter is, manufacturers can't get anything out the door that doesn't say "gaming, or, "RGB" on it. So, it's probably better just to skip using the RGB, and buy a board which will do the job you need it to do.

For example, I bought a low end Gigabyte B660 DDR-4 Intel board. It got terrible reviews from gaming sites. Something to the effect that, "well, that VRM won't hold up if you put 12700 in it and try to overclock it."

WTF do I care? (I said to myself). It's getting an i3-12100, and I'm sure it will do just fine with that. But, it still does have an RGB lane. There's no getting around that these days

 

[Avro Arrow]

Well, the fact of the matter is, manufacturers can't get anything out the door that doesn't say "gaming, or, "RGB" on it. So, it's probably better just to skip using the RGB, and buy a board which will do the job you need it to do.

For example, I bought a low end Gigabyte low end 660 DDR-4 Intel board. It got terrible reviews from gaming sites. Something to the effect that, "well, that VRM won't hold up if you put 12700 in it and try to overclock it."

WTF do I care? (I said to myself). It's getting an i3-12100, and I'm sure it will do just fine with that. But, it still does have an RGB lane. There's no getting around that these days

That's the thing. A lot of times, what they should do is just use the motherboard as it is and do whatever with the PC without worrying about how well it overclocks because most people don't do it. I've known how to OC for years but I only ever used it as a way to make a CPU live longer while I sought its replacement.

I'm like you in the way that to me, a motherboard either works or it doesn't. The rest is just background noise.

 

[pcnthuziast]

I avoid Gigabyte products like the plague.

 

[Bamda]

Since 2011 I have owned 2 Asus MBs and 2 Asus and 1 MSI GPU. I have had zero RMAs.

 

[captaincranky]

I avoid Gigabyte products like the plague.

I really wish you'd have given me a heads up on this before I built this lousy Gigabyte G-41 Intel rig, which I've used every day for the past 13 years.

 

[Avro Arrow]

I really wish you'd have given me a heads up on this before I built this lousy Gigabyte G-41 Intel rig, which I've used every day for the past 13 years.

Yeah, me too, eh? I have a Gigabyte 990FX motherboard that I bought in 2011 that still works. Anyone who "avoids whatever brand like the plague" because they think that it's crap is a fool. If any of these brands were crap, they wouldn't still exist.

Now, I do refuse to buy anything from MSi but that's because they treated me badly when one of their flagship motherboards failed on me after about 17 months, not because I think that their products are bad. The funny thing is that the cheap ECS/Elitegroup IC780M-A2 that I bought to replace the failed K9A2 Platinum still works to this day, 15 years later.

I use the Phenom II X4 940 to play a specific online game from 1998 that requires a GeForce card. I'm not going to bother using an 8500 GT in a modern motherboard so I just use that. It's funny because so many people used to say that ECS was garbage. It's amazing how people get these absurd thoughts in their heads and then use those thoughts as a guide.

 

[Endymio]

I love it when people clearly don't understand how the Steam survey works but act like they do. You don't have to be invited to participate.

Oops! Opt-in surveys are even less reliable than invitational ones. Neither should be relied upon, nor any survey which uses a non-random sample set.

 

[Endymio]

Random sampling works quite well, as long as you get a good sample size. According to this article on the subject, a sample size should be minimum 100, maximum 10% of the total population. They cap this at 1000 individuals

That article was written by someone who failed their stats class. Sample size is determined by your null hypothesis, the distribution of your data, and the degree of statistical significance required. If you're attempting to determine the percentage of the population that has an ultra-rare disease with five-sigma certainty, then sampling 1000 isn't nearly enough, whereas if you're attempting to prove that "most adults have had at least one cold in their life", then a sample size of a dozen is probably enough. The figures he stated are true only for binary outcomes on normal distributions (I.e. what percentage of people voted Democrat vs. Republican).

Edit: And, IIRC, the Steam Survey isn't a random sample anyway; it has an opt-in element, which skews the results.

 

[Avro Arrow]

You are actually wrong about this. Random sampling is a well-proven methodology in statistics, and it's actually explained fairly well in this reddit post about the July 2021 SHS:


Random sampling works quite well, as long as you get a good sample size. According to this article on the subject, a sample size should be minimum 100, maximum 10% of the total population. They cap this at 1000 individuals, and the SHS definitely surveys way more than that, so you can rest assured it can be expected to be accurate to within 3% up or down. And you know what, AMD cards have expanded a bit in the latest surveys anyway, so I'm happy enough lol

I'm not wrong because it's supposed to work and it doesn't. You can make all the excuses for them that you want to but it doesn't change that one simple truth.

 

[Avro Arrow]

I had quite a few Palit cards over the years, I stopped using Asus motherboards and switched to Supermicro. ... EVGA, silverstone for PSUs and crucial/micron for memory.

Motherboard brands that I've owned: Biostar, ASRock, MSi, ECS/Elitegroup and Gigabyte
GPU brands that I've owned: ATi, Albatron, Palit, XFX, Gigabyte, nVidia and Sapphire
PSU brands that I've owned: OCZ, Corsair, EVGA, Gigabyte (and a bunch of no-namers)

In general if you want something good make sure it hasn't got any RGB saying gaming on it but at the same time it's not the cheapest brand available.

Yep, I avoid RGB because I'd rather they spend more on the product and not the lighting. However, I ALWAYS buy the cheapest brand available when it comes to motherboards and video cards. It's something I had learnt when I worked for Tiger Direct because the differences between them are slight at worst and non-existent at best. My two XFX HD 4870s still work like the day I bought them and so too does my Palit 8500 GT. I threw out my Albatron FX-5200 and XFX 6200 because those were AGP and so became useless.

Out of all of those, only my MSi motherboard failed and it was a K9A2 Platinum, MSi's flagship AMD 790FX motherboard. It only lasted 15-16 months and so I said "I'm not paying top-dollar for a motherboard ever again!" and replaced it with a cheap ECS IC780M-A2. I gave it to my mother with my Phenom II X4 940 for her HTPC about eight years ago and now I have it again because I replaced it with an FX-8350 so that she could use Windows 10. That cheap as dirt board still works to this day.

As long as it's a brand that you're familiar with, you don't need to worry when it comes to motherboards and video cards. Now, I agree that PSUs are different but I've never had a no-name PSU fail on me. The most ironic thing that I've ever encountered in my 34½ years of building PCs is that when I worked at Tiger Direct, I learnt that having a good brand-name PSU is extremely important and I also learnt about the 80+ system. Back then there was 80+, 80+Bronze, 80+Silver and 80+Gold. No Platinum or Titanium yet.

So, armed with this new knowledge, I replaced whatever no-name PSU I was using with an OCZ GameXStream 1000W PSU because I was jumping from my Palit GeForce 8500 GT to an XFX Radeon HD 4870 and I wanted no problems. OCZ had bought PC Power and Cooling, a legendary PSU company, not long before and the GameXStream had a 3-year warranty so I figured "I should have no issues with this". I was wrong and the OCZ GameXStream, my first brand-name PSU, failed while I was attending university less than a year after purchase.

To their credit, OCZ replaced it with an OCZ Z1000, a 1000W 80+Gold PSU with a single 84A 12V rail, a far more expensive model than the GameXStream. However, that one didn't even last as long as my GameXStream did (I know, it was insane) and six months later, I had to RMA the Z1000. Happy ending though because OCZ replaced that one with an OCZ Z1000M which is the same as the Z1000 but fully modular. That one still works and I used it in a mining rig with an FX-8350, RX 5700 XT and RX 6800 XT for 6 months, running it 24/7. No problems, ever.

So, the first time I paid big for a motherboard, I got burnt. The first time I paid big for a PSU, I also got burnt. It doesn't matter what the brand is because the odds of something going wrong are astronomically small, no matter what brand you buy. Even the Biostar A320MH that I bought on clearance for $40 at Canada Computers works just fine. I didn't even need it, I just couldn't say no to a new AM4 motherboard for $40.

 

[Hodor]

The funny thing is that the ONLY motherboard that has ever failed on me was an MSi.

I never considered MSI a good brand. Even though it's advertised everywhere as a top brand, in the past I had so many problems with their products that for me MSI is a synonym for low-quality.

 

[captaincranky]

I never considered MSI a good brand. Even though it's advertised everywhere as a top brand, in the past I had so many problems with their products that for me MSI is a synonym for low-quality.

The 1st, last, and only piece I'll ever buy from MSI was a GT-730 card. I got it during the great video card shortage for $50.00, as I couldn't get an Asus GT-1030 which I was used to using. (not a gamer).

I came through with no DVI output, but had VGA. I thought it was me. It lasted about half an hour on the VGA output and black-screened.

So, we're back to, (the soon to be dearly missed), EVGA, and Asus for our VGA needs.

 

[Hodor]

The 1st, last, and only piece I'll ever buy from MSI was a GT-730 card. I got it during the great video card shortage for $50.00, as I couldn't get an Asus GT-1030 which I was used to using. (not a gamer).

I came through with no DVI output, but had VGA. I thought it was me. It lasted about half an hour on the VGA output and black-screened.

So, we're back to, (the soon to be dearly missed), EVGA, and Asus for our VGA needs.

I'm not surprised by this at all. I was more surprised seeing MSI as one of the top brands. Maybe they have advanced to medium level, but top..... nope.

 

[Avro Arrow]

I never considered MSI a good brand. Even though it's advertised everywhere as a top brand, in the past I had so many problems with their products that for me MSI is a synonym for low-quality.

The thing is that MSi IS a good brand. They're ALL good brands and that's why they still exist. Even Biostar, one of the worst (I still remember a hilariously bad video card they made), has an over 90% survival rate for their products. At this point, you're splitting hairs and I saw that when I worked at Tiger Direct.

Choosing a brand is like choosing which oil to put in your car. Noobs buy by brand and experts buy by spec.

 

[Hodor]

The thing is that MSi IS a good brand. They're ALL good brands and that's why they still exist. Even Biostar, one of the worst (I still remember a hilariously bad video card they made), has an over 90% survival rate for their products. At this point, you're splitting hairs and I saw that when I worked at Tiger Direct.

Choosing a brand is like choosing which oil to put in your car. Noobs buy by brand and experts buy by spec.

Maybe, but I never had good experience with MSI. I'd rather buy Gigabyte than MSI, even if Gigabyte is allegedly worse than MSI in this article.

 

[Avro Arrow]

Maybe, but I never had good experience with MSI. I'd rather buy Gigabyte than MSI, even if Gigabyte is allegedly worse than MSI in this article.

Gigabyte is a great name. I have a Gigabyte 990FX motherboard that is still in operation today that I bought back in 2010 or 2011. I used it for the first half of 2022 as a mining platform for my RX 5700 XT and RX 6800 XT. I was stuck using an R9 Fury in my main rig but hey, it was good enough that I wasn't suffering.

I ended up giving it to my mother with an FX-8350 because I was sick of seeing her trying to hold on to using a Phenom II X4 940 with Windows 7.

 

[captaincranky]

Choosing a brand is like choosing which oil to put in your car. Noobs buy by brand and experts buy by spec.

I hate to disagree with this, but I feel compelled to do so. If someone buys blindly, "by brand", then you're correct. However, if a brand has established their product by reputation, performance, and durability, then there's equal opportunity for both noobs and experts to buy with confidence.

However, even a highly regarded maker, may have specific product types they excel with. For me, it's Gigabyte for boards, Asus and EVGA for video cards, and ostensibly Antec, (gone from the market?), for PSUs.

As for MSI, as I said before, I bought one $50.00 video card from them, and it was defective. A lot of other factors could have been involved in the supply chain, but MSI gets the blame. So I shouldn't have said that MSI is a "bad brand", but I'm certainly justified in giving them a vote of, "no confidence", regarding further purchases.

As for Gigabyte, I'm using a G-31 / Pent E-2200 (Q4 / 07), at this very moment. While it's not setting any performance records, it's still kicking at 15 years old. Would I buy a PSU or VGA from them? Probably not.

Once upon a time, I sold Hi-fi equipment. Technics turntables sold like hotcakes and close to 100% of people were thrilled with them. However their receiver line never met spec, (as tested by Consumer Reports). If they rated a receiver at "55warrs / channel, you were lucky if it broke 50. Kenwood's amplifier product however, always exceeded spec. If they said it was 50 watts a side, you were more likely to get 55. JVC was pure junk. So, if you bought "Technics" throughout the line, you were screwed. Mix and match was a far better approach.

I could go on about my experiences, since I got employee discounts and manufacturer promo pricing, so I went through all of this equipment fairly frequently. Does this attach today? Dunno. But today's QC and automation have upped the overall likelihood of, "luck with getting a good one", is increased demonstrably with time.

So, if you buy exclusively by brand, that's very foolish. But, if you buy by brand reputation specific to a certain type of product, then, (IMHO), you're a well informed consumer..

 

[Hodor]

Gigabyte is a great name. I have a Gigabyte 990FX motherboard that is still in operation today that I bought back in 2010 or 2011. I used it for the first half of 2022 as a mining platform for my RX 5700 XT and RX 6800 XT. I was stuck using an R9 Fury in my main rig but hey, it was good enough that I wasn't suffering.

I ended up giving it to my mother with an FX-8350 because I was sick of seeing her trying to hold on to using a Phenom II X4 940 with Windows 7.

I had several Gigabyte motherboards (and probably some graphics cards too) over the years, can't remember having problems with them.

 

[Avro Arrow]

I had several Gigabyte motherboards (and probably some graphics cards too) over the years, can't remember having problems with them.

Yeah, that doesn't surprise me. These companies have been doing this for a very long time. I always appreciated Gigabyte's use of dual-BIOS and extra-thick PCB in their Ultra-Durable line of motherboards. Sure, I've never had a problem but it's nice to know that if I ever accidentally brick the BIOS, the motherboard isn't dead.