EXPO: New processors in iBooks!
No one noticed it and Apple, incredibly, did not give any publicity to the thing, but the real novelty is bigger than the new iBooks not in FireWire, nor in the video output n in the new colors, but in the processor. laptop from Cupertino in fact no longer beats the traditional PPC750 but its successor, the PPC750CX. Therefore, what our site had predicted since last May came true when, against what some of the most respected and venerated sites of ours supported, we began to put forward the hypothesis that the new chip would be used in the fall at the first revision of the low-end laptop. Hypothesis that we have advanced still not later than two days ago. It is surprising (but perhaps not too much, as you will understand before the end of this article) that to learn of the change it is necessary to comb through the technical documentation for developers made available today on the site reserved for them. Very little understanding is so little care in highlighting this structural change if we consider, on the other front, how much emphasis IBM had placed last spring to highlight its decidedly advanced characteristics compared to the "old" PPC 750. In reality wonder should cease if we consider that due to its intrinsic characteristics and the less generous second-level cache, the processor in use in the new iBooks cushions the effects of the speed bump which on balance is less generous than the pure MHz say. But let's go for order by listing Firstly, the advantages of the PPC 750CX. The chip, produced as mentioned by IBM, is smaller, less energy-hungry, less hot and less expensive. For this reason new technologies have been used, such as a layer of copper for heat dispersion and the elimination of connectors for external second-level caches. This last measure – as IBM explained during last June at the Embedded processor forum – reduced the number of pins (256 instead of 360 of 750) and the size. Now with the internal cache the PPC 750CX has a size of 42 square millimeters compared to 40 of the 750 without cache. The small size allows to produce more chips from a single silicon wafer and therefore to reduce costs. Unfortunately this technique and we come to the negative implications, it does not allow to employ more than 256 Kb of second level cache without significantly affecting costs and size. A PPC 750CX with 512 kb of second-level cache onboard would be 33% larger than the one with 256 kb and would cost 40% more. The reduced size of the cache and the impossibility of mounting an external cache therefore minimize the advantage in terms of the speed of the on-board cache (which runs at the same speed as the processor, instead of the half as in the current PPC 750) and that of the FPU. In fact, IBM points out in some technical documents, despite the fact that the 750CX beat the 750 from yesterday by the same level of cache, in everyday use the difference is reversed in favor of its predecessors. This is because normally a computer with G3 also has a second level cache of 512 or 1 MB which determines superior performances of about 5% compared to the CX which can only use, as mentioned, only 256 kb cache. This, in fact , means that the 466 MHz iBook is on balance in terms of pure speed as an iBook with PPC 750 from just over 430 and that an iBook from 366 as one from 340. The situation should change with the release of the PPC 750CXe . The chip, an evolution of the 750CX, will be released early next year and will have a .18 micron circuit and a refined production process that will allow it to run at higher speeds. According to IBM at speeds above 550 MHz the CXe also beats the current 750 running at the same speed, so a computer equipped with a 600 MHz PPC 750CX and 256 kb cache will be faster than one with a (hypothetical) current G3 with the same MHz and 512 kb cache. IBM sources report that the CXe should be released in batches of 550 to 750 MHz towards the end of the current year. These details let us suppose that the PPC 750CXe could be the reference chip for the 2001 PowerBooks. High-end laptops, if our predictions prove to be correct, could be released in February next year at speeds of 600 and 700 MHz unless Apple decides to adopt Motorola's G4 processors, but given the move made with the iBook the hypothesis appears to us rather unlikely at the moment.