Why is AMD so popular

[Abu Hajaar]

I’m building a new PC and don’t understand why the new AMD processors are so popular. Their specs and performance isn’t any better than intel and from what I’m seeing they seem slightly more expensive, and sold out.

[Scheneighnay]

İlk olarak temps tarafından gönderildi:

İlk olarak Monk tarafından gönderildi:

Problem there is, for price, the chips it should be compared to now is the 9900k/10700k which are faster at nearly same price (not that you can get a 5600X at msrp) while the 10600k has dropped to like £220.

The AMD 5600X CPU beats the i9 10900K in 1080p gaming according to the following Toms Hardware article.

If the 5600X beats the i9 10900K, I think it's a safe bet it also beats the i7 10700K as well.

Source: https://www.tomshardware.com/reviews/amd-ryzen-5-5600x-zen-3-review

I've seen claims that the new i5 does the same oddly enough. The 10900K might not be too solid for gaming.

[Illusion of Progress]

İlk olarak Scheneighnay tarafından gönderildi:

I've seen claims that the new i5 does the same oddly enough. The 10900K might not be too solid for gaming.

Not surprising, but it's not that the Core i9 isn't good for gaming. It's more that unless you are thread limited, most chips within the same lineup (outside comparing extremes like low end to high end, such as locked CPUs to K SKUs) tend to perform mostly similar than not for the majority of games. Clock speed differences between higher end models within the same lineup are there but small, and other than that, extra cores (and cache) are what the higher end models offer, and most modern Core i5/Ryzen 5 CPUs have enough cores/threads to where the difference with more will be either nonexistent, or small.

This is why Core i5s and Ryzen 5s are usually the best value buys regarding price and performance. The higher end is usually if you're willing to try and chase more longevity, need it for squeezing out more performance for a higher end GPU, and/or have highly parallelized workflows that benefit for the additional cores/threads.

[Monk]

The times when a 5600x wins us down to games like csgo which can make use of the chips large cache, as soon as you go beyond that, the performance drops noticeably.

As for

İlk olarak Scheneighnay tarafından gönderildi:

İlk olarak xSOSxHawkens tarafından gönderildi:

No, what he is saying is that he used a Vega without issue, and that he suspects your issues were just that, yours.

I cant personally validate monks own experiance, but I can validate monk as a long term helpfull member of the forum with a good record for technical accuracy even when we dont agree.

I *can* personally validate that the Vega 64 is by no means a bad card and that the experiance you claim here:






Is not the norm, even for a reference blower model. I personally own one and it benches out great and performs well, even with the stock cooler.

https://www.3dmark.com/fs/24444972

Not quite to top 100 (top 100 for all vega 64 starts at ~23,300 vs my 22,800) but getting quite close, and well within limits to be a top contender card within its product line. Benches out above most all 1080's and can get on the tails of the 1080ti in some rare cases.

You might also notice that with the custom fan curve, despite running a stock blower cooler, the GPU is only running 63c, so yeh, as Monk said the blowers can do a good job assuming you are willing to deal with fan noise.

I am looking forward to getting the Morpheus II installed (stuck waiting on thermal pads) and then updating the benchmarks. I am hopeful to break into the top 100 for all air cooled single vegas in 3dmark.

No, he's a jackass who thinks user error is the only cause of thermal throttling and POST failures.
But I guess if I forum warrior long enough my word becomes law.

It's not that I don't think faulty cards happen, hell I had 5 bad 1080ti ftw3's, it can happen, however, your issue as described sounds more like you messing it up than an issue with the cooler, only thing that can be is clamping pressure unless you had a faulty chip, at which point RMA is the fix, but, as you didnt mention it not even boosting and being way below stated speeds, that seems unlikely, you did mention messing about undervolting it and hsvi g issues many others did not, which, based on your attitude I drew said conclusion, that you were the problem, perhaps that was wrong of me.

The vega 64 was and us a very good card, the problem was, they speed wayyyy to high memory on it which caused it to cost a good £150 more than it could of with gddr5/x, interestingly, it's the opposite problem their new 6000 series cards have where they overclock like a bat out if hell but don't scale due to the memory bandwidth being too low.
which means nvidia keep the crown, a 6900XT not restricted by its memory at 2800MHz would likely beat a 3090 in alot of areas, sadly, it doesn't scale past 2300 really.

[Scheneighnay]

İlk olarak Illusion of Progress tarafından gönderildi:

İlk olarak Scheneighnay tarafından gönderildi:

I've seen claims that the new i5 does the same oddly enough. The 10900K might not be too solid for gaming.

Not surprising, but it's not that the Core i9 isn't good for gaming. It's more that unless you are thread limited, most chips within the same lineup (outside comparing extremes like low end to high end, such as locked CPUs to K SKUs) tend to perform mostly similar than not for the majority of games. Clock speed differences between higher end models within the same lineup are there but small, and other than that, extra cores (and cache) are what the higher end models offer, and most modern Core i5/Ryzen 5 CPUs have enough cores/threads to where the difference with more will be either nonexistent, or small.

This is why Core i5s and Ryzen 5s are usually the best value buys regarding price and performance. The higher end is usually if you're willing to try and chase more longevity, need it for squeezing out more performance for a higher end GPU, and/or have highly parallelized workflows that benefit for the additional cores/threads.

really what I've learned by replacing my Vega 64 with a 1650 Super is that low-bins can do a lot more than you think they can, so I've been feeling around for the perfect cost-effective sweetspot.

The RX 6500 could be a game changer if it doesn't take AMD another year to launch it like when they took their sweet time with the 5500.

[Scheneighnay]

İlk olarak Monk tarafından gönderildi:

The times when a 5600x wins us down to games like csgo which can make use of the chips large cache, as soon as you go beyond that, the performance drops noticeably.

As for

İlk olarak Scheneighnay tarafından gönderildi:

No, he's a jackass who thinks user error is the only cause of thermal throttling and POST failures.
But I guess if I forum warrior long enough my word becomes law.

It's not that I don't think faulty cards happen, hell I had 5 bad 1080ti ftw3's, it can happen, however, your issue as described sounds more like you messing it up than an issue with the cooler, only thing that can be is clamping pressure unless you had a faulty chip, at which point RMA is the fix, but, as you didnt mention it not even boosting and being way below stated speeds, that seems unlikely, you did mention messing about undervolting it and hsvi g issues many others did not, which, based on your attitude I drew said conclusion, that you were the problem, perhaps that was wrong of me.

The vega 64 was and us a very good card, the problem was, they speed wayyyy to high memory on it which caused it to cost a good £150 more than it could of with gddr5/x, interestingly, it's the opposite problem their new 6000 series cards have where they overclock like a bat out if hell but don't scale due to the memory bandwidth being too low.
which means nvidia keep the crown, a 6900XT not restricted by its memory at 2800MHz would likely beat a 3090 in alot of areas, sadly, it doesn't scale past 2300 really.

I have old benchmark results that I'm digging around for; iirc it performed above average but I need to check.

I do believe it was a faulty card, but part of my argument is that higher wattage means more points of failure and a higher chance of getting a faulty card.

Edit: Found the userbenchmark result I had saved- all it shows is CLim: 1630MHz, MLim: 945 MHz
77th Percentile of Vega 64s on userbenchmark.

The filename says it was with the undervolt set.

[Yamantaka]

İlk olarak temps tarafından gönderildi:

The AMD 5600X CPU beats the i9 10900K in 1080p gaming according to the following Toms Hardware article.

If the 5600X beats the i9 10900K, I think it's a safe bet it also beats the i7 10700K as well.

Source: https://www.tomshardware.com/reviews/amd-ryzen-5-5600x-zen-3-review

OC'd 10700k seems to fare quite well. Also, depending on where one lives prices seem to vary quite a bit. At least here 10700K went as low as a bit over 300 euros making it tempting option with 8 cores over 5600x's 6 cores. Similarly, 3700x and 10700k cost about the same, while later being better for gaming.

Just thinking with price:performance ratio in mind. Intel pushed prices down quite nicely thanks to welcomed competition.

https://www.youtube.com/watch?v=JSwe1cOkZbY&t=266s

[Mad Scientist]

İlk olarak Scheneighnay tarafından gönderildi:

Literally the only thing he's been saying is "if you've had bad experience with high-wattage GPUs it must mean you don't know what you're doing and not that high-wattage can be a bad thing"

They're a complete egotistical jackass with no ability to think critically for a second and defends their own biases to the death.

I feel sorry for anyone who takes their word seriously.

I've taken a look a few pages back just to ensure that I get better context and to point out a few issues. First, is I wouldn't name-call, I cannot guarantee they wont report you for the "jackass" comment. As silly as that would be to report for, either (they) discredit or move on. Your 'housefire' comment wasn't accurate either but that's below and is talked about, anyway:

Let me catch one issue I already found:

İlk olarak Scheneighnay tarafından gönderildi:

1 10 minute test in a controlled environment is absolutely not indicative of real-world performance.

İlk olarak dOBER tarafından gönderildi:

Explain please. A game never runs on full power draw whole time. This benchmark does which means its way more stress and heat then games ever could.

Depending on the game you can easily draw full power the entire time (per card may vary), also based upon settings, resolution and hz rate if the temps remain below the threshold. Synthetic testing does not give full credibility, and is why it's only part of troubleshooting for competent techs; as real world scenarios can and easily will replicate issues that do not always replicate under the load of a benchmark and depending on the cause can not replicate at all within benchmarks.

Another issue:

İlk olarak dOBER tarafından gönderildi:

Maybe at stock settings but who buys a expensive K cpu and runs it at stock.

Whoever wants to for any reason they see fit, they don't have to be an overclocker to appreciate an unlocked CPU and may decide later to OC if they wish. Others get them for the power compared to the locked units. It doesn't matter.

Really?:

İlk olarak Monk tarafından gönderildi:

and price us the biggest indication abd unless its in a laptop or server farm, power doesn't really matter.

For a lot of people it matters. Older houses, equipment on the same line; power isn't infinite on a given residential line nor an industrial line. To a lot of people not exceeding certain wattage is the difference between running fine, and tripping at least one breaker while pulling heavy loads. Some houses for example, if you use regular lights a breaker can trip when using a workstation and standard high energy use bulbs instead of LED low power consumption lights if they share the same electrical line. Power matters, price matters, performance matters. Maybe not to everyone, but largely it has mattered a lot as I had a lot of people with power related issues by drawing too much power; some had to get an electrician to install a higher capacity line. Afterall, the breakers are there to prevent fires.

I'd say this is not accurate as well:

İlk olarak Scheneighnay tarafından gönderildi:

I'm steering clear of RTX 30 neo-fermi-housefire-edition with how high the wattages are.

According to Toms Hardware - as much as I question them sometimes (a history no one here will know about):

GPU Core
https://cdn.mos.cms.futurecdn.net/XH75XnNwKgtX72fcbAH4AH-970-80.png

Fan Speeds (room to push more):
https://cdn.mos.cms.futurecdn.net/WtiyieskpkwDturmTberfU-970-80.png

Fermis ran VERY HOT especially the 480, 570 and 580s. I'm talking 80C was entirely normal for them, that kind of hot. Seeing 90-92C was not at all out of the question either under heavy loads - usually more so the 580's. Yes, many peoples cards ran at this temperature and it stayed on the entire time.

İlk olarak dOBER tarafından gönderildi:

You clearly dont know what you are talking about. For example 3080 FTW3 Ultra with 450w xoc bios 2100mhz OC has below 50°C while running an extrem gpu benchmark like timespy. If you call this housefire i dont know how cold your card runs!

My 3080 450w 48°C
https://www.3dmark.com/spy/15901828

My 3090 450w bios was a little higher with 58°C
https://www.3dmark.com/spy/17410642

Your test. This does not define it for everyone
Also note if you're using an FTW3 Ultra - That's an EVGA unit and their higher end models always have custom cooling designs, and are usually not reference PCBs which Scheneighnay appears to be talking about. Of course EVGA units are going to run far cooler, especially if you're using Precision. If you're going to try to say "but mine" and you're using a board partner/vendor, with custom cooling and usually far superior cooling, that doesn't automatically make the other person not know what they're talking about, even reviewers had far higher temps than you, because they weren't using a custom model. Your 3090 is more than likely the triple fan non reference design as well, correct? Or is it the mostly reference design with a blower instead of a regular fan?

In the future, keep similar cooler units as their own readings for temps, rather than clustering custom units against reference units which again will vastly differ in temps, and often potential which is what the FTWs are literally designed for other than Kingpin units.


That should make you happy, Scheneighnay , even with me disagreeing about one of your comments. Now most of you are on a leveled playing field, so to say.

[Monk]

My 3090 FE overclocked to run between 2050-2150MHz with the fabs ramped up stays around 50c, they really aren't that hot if you don't mind a bit of fan noise.

Still waiting for my 3090 strix oc to see how that compares, but, I've a feeling it may not boost as high :(

[DeadBeat]

İlk olarak Yamantaka tarafından gönderildi:

OC'd 10700k seems to fare quite well. Also, depending on where one lives prices seem to vary quite a bit. At least here 10700K went as low as a bit over 300 euros making it tempting option with 8 cores over 5600x's 6 cores. Similarly, 3700x and 10700k cost about the same, while later being better for gaming.

Just thinking with price:performance ratio in mind. Intel pushed prices down quite nicely thanks to welcomed competition.

I just picked up a 10700k for the g/f and it was on sale for $500 Cdn. I happened to look at the site where I bought it today and now the intel chips are listed at much lower prices. 10700k was $580 (before sale) and is now $460(without sale). 10900k was $750 and is now $600.

[hawkeye]

İlk olarak DeadBeat tarafından gönderildi:

İlk olarak Yamantaka tarafından gönderildi:

OC'd 10700k seems to fare quite well. Also, depending on where one lives prices seem to vary quite a bit. At least here 10700K went as low as a bit over 300 euros making it tempting option with 8 cores over 5600x's 6 cores. Similarly, 3700x and 10700k cost about the same, while later being better for gaming.

Just thinking with price:performance ratio in mind. Intel pushed prices down quite nicely thanks to welcomed competition.

I just picked up a 10700k for the g/f and it was on sale for $500 Cdn. I happened to look at the site where I bought it today and now the intel chips are listed at much lower prices. 10700k was $580 (before sale) and is now $460(without sale). 10900k was $750 and is now $600.

11th gen is almost here, so they may be trying to clear inventory.

[DeadBeat]

İlk olarak hawkeye tarafından gönderildi:

11th gen is almost here, so they may be trying to clear inventory.

That's probably the case as there are quite a few intel procs still in stock whereas there are no AMD 5000s to be had at the retailer I deal with.