Discussion What we used to see VS. What we now see in reviews.

[bgunner]

Back in the 2000's and in the 2010's I think we had a better understanding of how GPU's sat performance wise in a given series. Review sites used to be able to afford or with help from the manufactures, get there hands on different brands of the same GPU's to do side by side comparisons. This led to performance differences usually in under 10FPS difference between them but there were times where one's cooling solution out performed another's allowing for quieter operation.

As of late these side by side comparisons have been few and far between. This I believe has led to a few regrettable GPU purchases by consumers where they have a bunch of options to choose from but don't really know how one compares to another. As history has proven the performance delta between different manufactures is small but each one is slightly different in there own ways. One may cool better at a lower noise level than the others but you may have one that is a little noisier and have better over all performance.

Now one consumer may be looking for the fastest GPU they can get in there price range and not care that the card is a little noisier because they want the raw power for framerates where as another consumer may want the quietest they can get because, lets say, they have the tower on top of the desk 2-3 feet from there head and they don't want to listen to it while gaming. This is where these side by side comparisons come in handy for the consumer.

Bouncing around from reviewer to reviewer only gives some of the picture since there test systems, methodology and test equipment are different. With this being said it is like comparing apples to apples but more like taking a Granny Smith apple from one state and comparing it to another Granny Smith that was grown on the other side of the country where soil, weather and pollutants can change the taste. They are comparable but not as accurately and leaves subjectiveness to the taste buds. One reviewer being able to test the same GPU from multiple manufactures gives a direct comparison on the same equipment so that there is no subjectiveness there. There is empirical proof where each one sits and each of there short comings along with the pros that they have.

This is not meant as a dig at any reviewer or company as I am a long time reader and lurker/helper here at the Tom's forums and site. This is just a comparison of how the times have changed what what I personally see as differences over this time span with reviews on GPU's. I understand that the GPU's that any reviewer tests are paid for by the company or once in a while cherry picked by the manufacturer and given to a review site for testing. There is a financial toll to test multiple GPU's of the same kind just from different company's and Review sites need to make money to keep up to date on equipment and software. Thus is the nature of this business.

I'd love to hear anyone's thoughts on this subject and what they look for when they are looking to purchase a new GPU.

 

[hotaru.hino]

A few things happened between then and now:

• GPUs operated at a fixed clock clock speed, maybe sometimes lowering for power savings, but for the most part, the clock speed listed on the box was the clock speed you got. These days, GPUs will clock themselves as fast as they possibly can within their thermal and power limits. The number on the box is mostly to satisfy some legal requirement (probably) and the GPU more of then than not exceeds this.
  
  
• Similarly most of the coolers back then were using the reference design and the only thing AIBs were doing was slapping their sticker on it. I have an EVGA 8800 GTS 512MB card and the only thing that's different about the stock card's cooler is EVGA put a decal on it. Options were also limited on how you can adjust the fan curve, if you could even do it.
  
  These days practically all AIBs have a "zero fan" mode when idling, so that end of the spectrum is pointless to compare. The noise at which the fans start working on may be of some importance and while some sites do compare noise levels (https://www.techpowerup.com/review/asus-geforce-rtx-4080-super-strix/39.html), it may not matter as much because you have other things in your case that are likely just as loud, if not louder, during those situations. Also now most video cards you can fully adjust their fan curves using something like MSI Afterburner or Fan Control.
  
  
• If we're talking about cooling performance, then sure, some sites still do that for stock configuration, but there are plenty of ways to adjust how the video card works to maintain a semblance of being relatively cool

Basically, if you're the kind of person who cares a lot about fan noise, temperatures, etc., you're probably going to be the person who's going to tweak the snot out of the card. And sure, while you might get a better baseline from a card that performs better on stock settings, chances are, getting a technically worse card isn't going to affect the actual performance of the card all that much in the grand scheme of things.

For everyone else who just wants a card to plug in and play, most cards are built such that people can use them and go about their merry way.

 

[Gururu]

Following GPU reviews closely for about 1 year now, I think they are usually about as effective as they can be. The market is saturated with just too much of the same. It is just pure lunacy that there is a product for every $50 increment. I feel like nVidia opened the door for this when they started pushing their chips on third parties. It was only inevitable that they would create so many products so these parties could then distinguish themselves and compete.
So reviewers dont have a lot to offer in terms of discovery. The next hot product is some variation of an existing product. It’s a darned quagmire of video cards. The good news is that reviewers will have dozens of reviews for each product cycle.

 

[Joseph_138]

Try buying a dozen 4090's for a review video. We'll wait. 🤣

 

[artk2219]

Try buying a dozen 4090's for a review video. We'll wait. 🤣

Man, you just made me realize I could buy 12 rx 6600's or Arc A750s for the price of one RTX 4090. I haven't looked at performance per dollar in a while.

 

[hotaru.hino]

I would say the problem of "too many cards to review" existed by around 2003. NVIDIA was already in the AIB game and I think ATi at that point opened their doors to AIBs.

A lot of the players you see on the market today have existed for a long time. And if you consider how many of them exist...

 

[bgunner]

Try buying a dozen 4090's for a review video. We'll wait. 🤣

And this is why I said and I Quote:

I understand that the GPU's that any reviewer tests are paid for by the company or once in a while cherry picked by the manufacturer and given to a review site for testing. There is a financial toll to test multiple GPU's of the same kind just from different company's and Review sites need to make money to keep up to date on equipment and software.

I Understand that testing mutli GPU's of the same kind, especially the obscene priced highest tear, GPU's would be extremely expensive. Only as of late has the price of these GPU's gone into the stratosphere and more so after covid hit where people kept buying GPU's from scalpers at absurd prices which proved to company's that the public would pay that price for there offerings. This is, in a way, a self inflicted wound.

 

[bgunner]

I would say the problem of "too many cards to review" existed by around 2003. NVIDIA was already in the AIB game and I think ATi at that point opened their doors to AIBs.

A lot of the players you see on the market today have existed for a long time. And if you consider how many of them exist...

I absolutely see your point with your statement. There are far to many AIB's out there for one company to cover them all where in the past each reviewer was able to cover 3-4 GPU's. At this point AIB's sent out the same GPU to multiple reviewers so some reviewers would cover the same GPU but have some that were different than the others. This is where we could jump site to site, compare the same exact GPU and have a better idea where the others tested sat on a different sites comparison.

This lead to the consumers having a better understanding of where the GPU they were considering sat in noise, performance, power usage, overclocking ect..