The article says that although Ryzen does outperform Intel's offerings, the deal breaker will be its price. If it's too expensive, it won't sell well. If it's just the right price, it will be an overnight success.
The article is spot on in what is essentially the main performance indicators: price, and the price/performance ratio.
Ryzen's performance means close to nothing if its priced out of the market. Nowadays, most computational needs are already well met and even surpassed by what's on the market for the past, say, half a dozen years. AMD's newest product may blow Intel's offerings out of the water, but that means close to nothing as they already outperformed the market's needs. In a world where the main incentive to increase performance is laid squarely on e-penis bragging rights, either the price makes sense or it's yet another nice-to-have-but-don't-bother offering.
I would not say that performance has been met. Performance has been met for those who pay a lot of money and then only use their computer for Chrome or photoshop. It still performs quite poorly for many other things, like compiling, physics calculations, network tweening in multiplayer games, and all the things you can do with your computer outside of the basics. Thats not e-penis, thats real work/applications that still suffer quite badly, even with an I7 extreme edition.
Personally I've been dissapointed in how CPU's haven't improved in the same way GPU's have. I'd much rather that CPU's start getting bigger rather than smaller, and put more effort into single-core performance.
> It still performs quite poorly for many other things, like compiling, physics calculations, network tweening in multiplayer games, and all the things you can do with your computer outside of the basics. Thats not e-penis, thats real work/applications that still suffer quite badly, even with an I7 extreme edition.
The market share associated with those use cases is between very slim and practically nonexistent.
Then, you've isolated the market segment which is willing to pay a premium for performance, and then argue that cost is a factor.
> Thats not e-penis, thats real work/applications that still suffer quite badly, even with an I7 extreme edition.
Quite frankly, those who care for performance and are bound to a budget are also not waiting for new product releases. They are in fact paying attention to the used parts market, where nowadays it's possible to buy dozens of server processors for chump change.
As an example, nowadays it's possible to buy on ebay dozens of bulldozer-based Opteron processors for the same amount of cash that a brand new i7 processor is sold for, and whoever is interested in computational power can easily setup a 4-socket Opteron server for less than €1000.
>The market share associated with those use cases is between very slim and practically nonexistent.
If so, then why do extreme editions and the 6700K exist? Those are significantly slimmer in usage than dual core I7's, I5's or I3's. In reality, there are a lot of power users who work with video, code, and plenty of other tasks that demand the highest computing power. These are usually also backed by a business, so they can charge a good amount for it.
>Quite frankly, those who care for performance and are bound to a budget are also not waiting for new product releases. They are in fact paying attention to the used parts market, where nowadays it's possible to buy dozens of server processors for chump change.
That depends upon your budget. If you're working proffessionally with the things I've listed there, chances are an $800 cost is tiny compared to the returns of a productivity increase.
>As an example, nowadays it's possible to buy on ebay dozens of bulldozer-based Opteron processors for the same amount of cash that a brand new i7 processor is sold for, and whoever is interested in computational power can easily setup a 4-socket Opteron server for less than €1000.
That statement largely ignores all the aspects of proffessional machines, and single-core performance. Getting a bunch of old opterons and slapping them in a machine is not a reliable setup. Even if it was, there are plenty of things that cannot be threaded well, its just how some algorithms work. If everything was threaded, Intel could have shoved 12 cores into all their chips long ago.
I get the vibe that the author thinks that AMD will just launch a 8c/16t processor.
But AMD will launch an entire lineup, from the 8c/16t high end processor to at least a 4c/4t model.
That said, 500$ for a 8c/16t CPU seems reasonable. The 6c/12th model must be around 350$ to be competitive and a 4c/8t CPU will own the market at a 200-250$ price point. The 4c/4t model must be below 200$ to compete with Intels i5 lineup.
I'm not sure if AMD has a response to the recent release of the Pentium G4560. For 64$ it offers the same performance we expected from a i3.
7 comments
[ 1.5 ms ] story [ 35.5 ms ] threadThe article says that although Ryzen does outperform Intel's offerings, the deal breaker will be its price. If it's too expensive, it won't sell well. If it's just the right price, it will be an overnight success.
In short, platitudes.
Ryzen's performance means close to nothing if its priced out of the market. Nowadays, most computational needs are already well met and even surpassed by what's on the market for the past, say, half a dozen years. AMD's newest product may blow Intel's offerings out of the water, but that means close to nothing as they already outperformed the market's needs. In a world where the main incentive to increase performance is laid squarely on e-penis bragging rights, either the price makes sense or it's yet another nice-to-have-but-don't-bother offering.
Personally I've been dissapointed in how CPU's haven't improved in the same way GPU's have. I'd much rather that CPU's start getting bigger rather than smaller, and put more effort into single-core performance.
The market share associated with those use cases is between very slim and practically nonexistent.
Then, you've isolated the market segment which is willing to pay a premium for performance, and then argue that cost is a factor.
> Thats not e-penis, thats real work/applications that still suffer quite badly, even with an I7 extreme edition.
Quite frankly, those who care for performance and are bound to a budget are also not waiting for new product releases. They are in fact paying attention to the used parts market, where nowadays it's possible to buy dozens of server processors for chump change.
As an example, nowadays it's possible to buy on ebay dozens of bulldozer-based Opteron processors for the same amount of cash that a brand new i7 processor is sold for, and whoever is interested in computational power can easily setup a 4-socket Opteron server for less than €1000.
If so, then why do extreme editions and the 6700K exist? Those are significantly slimmer in usage than dual core I7's, I5's or I3's. In reality, there are a lot of power users who work with video, code, and plenty of other tasks that demand the highest computing power. These are usually also backed by a business, so they can charge a good amount for it.
>Quite frankly, those who care for performance and are bound to a budget are also not waiting for new product releases. They are in fact paying attention to the used parts market, where nowadays it's possible to buy dozens of server processors for chump change.
That depends upon your budget. If you're working proffessionally with the things I've listed there, chances are an $800 cost is tiny compared to the returns of a productivity increase.
>As an example, nowadays it's possible to buy on ebay dozens of bulldozer-based Opteron processors for the same amount of cash that a brand new i7 processor is sold for, and whoever is interested in computational power can easily setup a 4-socket Opteron server for less than €1000.
That statement largely ignores all the aspects of proffessional machines, and single-core performance. Getting a bunch of old opterons and slapping them in a machine is not a reliable setup. Even if it was, there are plenty of things that cannot be threaded well, its just how some algorithms work. If everything was threaded, Intel could have shoved 12 cores into all their chips long ago.
But AMD will launch an entire lineup, from the 8c/16t high end processor to at least a 4c/4t model.
That said, 500$ for a 8c/16t CPU seems reasonable. The 6c/12th model must be around 350$ to be competitive and a 4c/8t CPU will own the market at a 200-250$ price point. The 4c/4t model must be below 200$ to compete with Intels i5 lineup.
I'm not sure if AMD has a response to the recent release of the Pentium G4560. For 64$ it offers the same performance we expected from a i3.