Report: AMD Readying FX-8770 and FX-9000 CPUs

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.

AMD is having a hard time replicating Intel successes... maybe they have decided to take a shot at duplicating Intel's failures for a change.

I have one P4 space-heater on my shelf and I definitely would not want another... that thing gets annoyingly loud. (Stock HSF in an Antec Aria mATX SFF - worst combination ever.)
 
i couldn't be disappointed any more....the TDP is higher and everything is the same but the clock base have increased just 0.8ghz !i'm betting it's atleast 140$ more expensive than the FX 8350...so why would you buy a FX 9000?don't!buy a CPU cooler instead and over clock your 8350 it would be the same as the FX 9000 both by price and performance...
 
"Nothing more than a factory OC AMD FX-8350 out of the box. AMD FX-8350 with the right cooling can hit 5Ghz no problem. It might match intels 4770k factory but the FX-9000 has no head room to overclock at all. the 4770k has a lot of head room. "

I really disagree with this comment. I'm no AMD nor Intel fan boy and I think both companies makes their good and not so good stuff. If I want an inexpensive system that can do basic thing then I'd go with AMD every time. If I'm gamer them depending on what games I play I could go with either. Some games favor Intel and some favor AMD, but both companies have CPUs that can push the most powerful video at playable frame rates. Now if I was doing more than just games like video or audio encoding I'd actually pick Intel for that because their hex cores with hyper threading tends to beat out AMD's modular core system. I think that much wattage is ridiculous for any CPU, but to say that the i7 4770k has more headroom is very uneducated. The 4770k unfortunately has the same design flaw as the 3770k does, rather than using a soldered on cover for the CPU like they did with the 2600k, 2700k, or even the i5 2500k what they did was between the cover and the CPU they use a real cheap ISP(not something like Artic Silver 5 at least) and because of this it really limits the potential overclocking ability these CPUs should have. There's a couple youtube videos that demonstrate this flaw and also show how do modify, repaste and lower your idle and load temperatures by doing these things. Even overclocking FX-8350 you'll still encounter heat issues and probably have wattage use that high as well if not possibly higher. I have nothing against overclocking, but wonder why sometimes why would you want to push your CPU higher if you already have a good one with a decent video card, a good ssd and good fast ddr3 that already max out your games at more than playable frame rates? That's just a waste of electricity, money that could be better spent and possibly risking your parts to go bad sooner. But everyone's entitle to their own opinion, so let just agree to disagree :).

THX.
 

Someone did a de-lidding experiment and found out in the process that the package thickness from the bottom of the CPU substate to the IHS top minus the IHS thickness indicates a 0.06mm (60 micron) gap between the IHS and CPU die caused by the IHS' glue.

Removing the glue, replacing the thermal compound by known material and emulating the glue gap using paper shims to reproduce the same total package thickness caused other pastes to produce WORSE temperatures than Intel's paste, indicating that Intel's IHS paste is actually on par with the best stuff on the market.

Removing the shims reduced core temperatures by as much as 30C during overclocking.

The gap between IHS and CPU die appears to be the major problem here, not the TIM material. The questions is: is this excessively thick gap accidental or by design? And if by design, why?

The reasons for ditching solder for paste seems simple enough: avoid uBGA failures during the IHS attachment process by eliminating soldering and reduce mechanical stresses on the core/uBGA from thermal cycling, external forces such as HSF during mounting and shipping by using soft material between IHS and the die to let the IHS direct external forces away from the die and delicate uBGA connections.
 


they stopped soldering (i would assume) as locked chips do not need soldering as ivy bridge runs fine stock speeds. if they were to solder the k chips and not the locked chips, it would complicate the manufacturing process
 
Thanks for helping to buyers.
5.jpg
 
"The gap between IHS and CPU die appears to be the major problem here, not the TIM material. The questions is: is this excessively thick gap accidental or by design? And if by design, why?"

That does make sense. If you think about it even with a decent paste if you have that much of a gap your conductivity isn't going to be the best. The people that got the best OC result were the ones that not only removed the cover and repasted it, but also got rid of the glue as well. I do wonder why Intel never addressed that complaint, but my guess would be if sales are going real well then I really don't think they'd care. My guess would be it's about lowering their manufacturing costs and possibly forcing people looking for the best performance to upgrade to the newest chip sets and CPUs. I also noticed that Intel does tend to raise the price of their old CPUs when they release a new chip set and CPU series and I remember when they used to lower their prices, not anymore. If you can't overclock that well then you really couldn't expand the lifetime of your CPU to compete with the new CPUs. This is why I'd glad I never upgraded after getting the i7 2600k. As I said I'm not a big fan of overclocking, but when my chip's warrantee expires I know I could easily push mine to 4.6 on a good air system or maybe 4.8 or even 4.9 with liquid cooling. The next chip I'd even get from Intel would be a hex core with hyperthread. What I have already can max out my 1080p monitor on almost every game I've played at 1920 by 1080 with a GeForce GTX 570 and 16gigs of memory. I don't care for the idea of getting a new system every 2 years, I'd like to get at least 4 to 5 years out of my CPU and at least 2 and half to 3 years from my video card. With the way technology moves that gets harder to do 🙁.

THX for the feedback.
 

A mid-range Intel CPU used to cost ~$500. Not anymore with i5s gravitating around the $200 and i7 around the $300 price points.

Intel used to discontinue mainstream chips around the $193 price point instead of offering lower-cost parts. Today, Intel's price range extends all the way down to ~$40 which I'm not sure even covers costs - last time I read about manufacturing costs, it was estimated at around $80 per CPU and with fab costs doubling every few years while mainstream x86 die sizes remain relatively constant, I doubt manufacturing costs drop much. I'm guessing the Celerons and lowest-end Pentiums are made mostly from dies that would otherwise have been complete write-offs to reduce net loss on those.

Intel used to only have two mainstream CPU product lines. Now they have five. Makes it hard to lower prices any further without price listings starting to trip all over each other.

As for why Intel "increases" their prices on old chips, they don't. What happens is Intel stops production, distributors hoard the last of Intel's leftovers and once distributors and retailers start running out of stock, they raise their prices on the few remaining high-demand SKUs to either squeeze more revenue out of their high-demand stock (their reward for taking a chance with stocking those chips) or steer buyers towards less popular parts that may also still be in stock and need to be cleared out of inventory. Intel has little to no loose inventory to manipulate market prices with at this point; the market charges whatever it wants.
 
" the market charges whatever it wants."

I remember a saying people have for that when I took Ecomonics in high school (which I failed BTW and I wish I had listened better) was "supply and demand". That would be a smart strategy for a retailer to horde or keep a overstock right before a new chipset and CPU series comes out. I know some people that still run the old LGA 775 dual core and I know when they looked around to upgrade to a quad core it really isn't worth it. I just went to pricewatch.com which is a good site to find older parts and the Q9550(which was a great CPU for the time and that what I upgraded from) was $170 for an OEM with no warrantee or HSF at the lowest price point and I'm not even sure if that would be the newest revision of that chip that OCed the best. I know when I auctioned mine on Ebay I got $200 for the newest revision by the end of the auction and that's more than I paid for the chip brand new lol lol.

THX again.
 
i really doubt any that any chip out there would cost more then 10 bucs to manufacture, and i am more than sure thant no single chip costs 80 bucs to manufacture per piece, that would only be resonable if that would include all the development procces aswell
 
i really doubt any that any chip out there would cost more then 10 bucks to manufacture, and i am more than sure thant no single chip costs 80 bucs to manufacture per piece, that would only be resonable if that would include all the development procces aswell
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom